Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
In fungi, success of mating requires that both cells agglutinate, modify their extracellular envelopes, and fuse their plasma membranes and nuclei to produce a zygote. Here we studied the role of the Schizosaccharomyces pombe Dni1 protein in the cell fusion step of mating. Dni1p is a tetraspan protein bearing a conserved cystein motif similar to that present in fungal claudin-related proteins. Dni1p expression is induced during mating and Dni1p concentrates as discrete patches at the cell–cell contact area and along the mating bridge. Proper Dni1p localization depends on Fus1p, actin and integrity of lipid rafts. In dni1Δ mutants, cell differentiation and agglutination are as efficient as in the wild-type strain, but cell fusion is significantly reduced at temperatures above 25°C. We found that the defect in cell fusion was not associated with an altered cytoskeleton, with an abnormal distribution of Fus1p, or with a defect in calcium accumulation, but with a severe disorganization of the plasma membrane and cell wall at the area of cell–cell contact. These results show that Dni1p plays a relevant role in co-ordinating membrane organization and cell wall remodelling during mating, a function that has not been described for other proteins in the fission yeast.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In S. pombe, heterothallic strains belong to one of two specialized cell types: M (h- cells) or P (h+ cells). Additionally, there is a homothallic strain (h90) that changes its mating type during proliferation (see Arcangioli and Thon, 2004). Mixed populations of h- and h+ cells can proliferate actively in rich medium without conjugating. In the absence of nitrogen, the level of cAMP decreases in the cell, which results in the dissociation of a heterotrimeric G protein and the ste11+-mediated expression of many genes involved in sexual development (Yamamoto et al., 1997; Davey, 1998; Nielsen, 2004). Then, cells produce mating pheromones that bind to specific receptors present in the membrane of the cells belonging to the opposite mating type and initiate the mating process by differentiating into specialized cells called shmoos. Shmoos are G1-arrested cells that grow in a polarized way to facilitate the contact with a shmoo from the other mating type (Nielsen, 2004). Then, the Mam3p and Map4p agglutinins (Yamamoto et al., 1997; Mata and Bahler, 2006; Sharifmoghadam et al., 2006) facilitate and strengthen the union between the cells (Calleja and Johnson, 1971; Calleja, 1987).
When the union between the parental cells is stable, the cell walls that separate them degrade and the plasma membranes fuse, allowing diffusion of the cytoplasmic contents between both cells (Calleja et al., 1977a). This step of the mating process has been subjected to detailed microscopic analyses (Streiblová and Wolf, 1975; Calleja et al., 1977a,b) but has not been characterized at the molecular level. It is known that a functional cytoskeleton is required for this process, since mutations in cdc3+ or cdc8+ (coding for profilin and tropomyosin respectively), and treatment with microtubule-destabilizing agents lead to a defect in cell fusion (Petersen et al., 1998a,b; Kurahashi et al., 2002). Cell fusion also requires Fus1p and Cfr1p. fus1Δ and cfr1Δ mutants have no apparent defects during vegetative growth, but they show a defect in cell fusion. Fus1p is a formin homologue that localizes to the tip of the shmoo and is required for the organization and stabilization of F-actin at the projection tip (Petersen et al., 1995; 1998b,c). Cfr1p is a Golgi protein that regulates mating through a Fus1p-independent pathway (Cartagena-Lirola et al., 2006). Once the cells fuse, karyogamy takes place, producing a diploid zygote. In S. pombe, diploids are unstable so they undergo meiosis immediately unless they are kept in rich medium. When meiosis is complete, four haploid spores develop (Yamamoto et al., 1997; Davey, 1998; Nielsen, 2004).
Since the cells are only able to initiate mating in a nitrogen-scarce medium, systematic studies have been undertaken using this condition to identify the genes that participate in this process (Mata et al., 2002; Mata and Bahler, 2006). In one of these studies, the genes that were induced in minimal medium without nitrogen were grouped in clusters according to their time of expression (Mata et al., 2002). fus1+ belongs to the cluster denoted as ‘Genes induced in response to nitrogen starvation (delayed)’, which includes several genes involved in the production of pheromones and their receptors. To gain information about the cell fusion step of mating, we are studying some of the genes included in that cluster. In a previous work we characterized the SPBC21D10.06C ORF, which turned out to code for the h+-specific agglutinin Map4p (Sharifmoghadam et al., 2006). Here, we describe the characterization of dni1+ (from delayed minus-nitrogen induction, corresponding to the SPAC31G5.07 ORF).
According to computer-assisted predictions, Dni1p is a tetraspan (a protein with four transmembrane domains). There are different families of tetraspans that share a common topology and structure, with some residues conserved at specific positions of the extracellular loops, although there are also atypical members of each family. Dni1p is most similar to some fungal members of the claudin-stargazing family of proteins, which include the Sur7-related and the Fig1-related proteins (Muller et al., 2003; Walther et al., 2006; Zhang et al., 2006; Alvarez et al., 2008; Grossmann et al., 2008) (see Fig. 1 and Fig. S1). The Sur7-related proteins have been implicated in protein turnover because they are associated with eisosomes, which are endocytic sites at the plasma membrane, and with the MCC domains (ergosterol-rich Membrane Compartments of Can1), which are sites where the proteins are protected from endocytosis (Walther et al., 2006; Alvarez et al., 2008; Grossmann et al., 2008). Fig1p-related proteins participate in the Low-Affinity Ca++ influx System (LACS) and are required for proper cell fusion during mating in S. cerevisiae (Erdman et al., 1998; Muller et al., 2003; Aguilar et al., 2007), and for polarized growth in Candida albicans (Brand et al., 2007). Dni1p shares 21% identity and 45% similarity with Fig1p. Since LACS activation only occurs in nitrogen-rich media (Muller et al., 2001), and since S. pombe cells do not mate under these conditions (Yamamoto et al., 1997; Davey, 1998; Nielsen, 2004), we wanted to study whether Dni1p played any role in calcium uptake and/or cell fusion during mating. We found that Dni1p localized as a discrete patch to the tip of the shmoo. dni1Δ null mutants showed a temperature-dependent defect in cell fusion that is not aggravated by the deletion of dni2+ (a second S. pombe Fig1p homologue). Genetic and microscopic analyses revealed that Dni1p and Fus1p are functionally related, Fus1p being necessary for the correct localization of Dni1p to the tip of the shmoo. We also found that Dni1p was required for the correct organization of the plasma membrane and the co-ordination between membrane and cell wall remodelling at the area of cell–cell contact.
Dni1p belongs to the fungal family of claudin-stargazing proteins
According to the data existing in the databases, the SPAC31G5.07 ORF is induced during the initial moments of mating and codes for a protein that shares sequence similarity with the S. cerevisiae Fig1p. A closest sequence analysis allowed us to find out that the SPBC4.01 ORF codes for a second FIG1 homologue that is also expressed in response to nitrogen deprivation (Mata et al., 2002). We have termed these genes dni1+ and dni2+ respectively. Figure 1 shows a sequence alignment between S. cerevisiae Fig1p and Sur7p, S. pombe Dni1p and Dni2p, C. albicans Sur7p, and human Claudin1. The alignment shows the conserved amino acids in the Cys-containing GΦΦGxC(8–20 aa)C motif, where Φ = Y, F, L or M. This motif is characteristic of the Fig1-family proteins and is related to the WxxW/YxxC(7–10 aa)C motif present in the Sur7-family proteins and to the GLWxxC(8–10 aa)C motif present in claudins (Zhang et al., 2006; Alvarez et al., 2008). The alignment indicates that Dni1p and Dni2p belong to the family of fungal claudin-stargazing proteins.
Dni1p levels increase in response to nitrogen starvation and Dni1p localizes to discrete patches at the tip of shmoos
The results obtained from genome-wide analyses showed that dni1+ was upregulated in response to pheromones or nitrogen deprivation (Mata et al., 2002; Mata and Bahler, 2006). In order to confirm that Dni1p was involved in mating, we analysed the level of this protein during this process by Western blot. We found that the amount of Dni1p (pointed by an arrow in Fig. 2A) increased after 2 h of incubation in the absence of nitrogen, reaching a peak after 4–6 h of incubation. Thereafter, the Dni1p signal decreased and new bands that might correspond to degradation products and the GFP alone could be observed (asterisks in Fig. 2A). Probably, these bands were produced because the Dni1 portion of Dni1–GFP was degraded after the initial steps of mating releasing the GFP portion. The maximum levels of Dni1p were reached when cells were initiating mating, and before the sporulation efficiency was significant. These results suggested that Dni1p might have a function in mating.
To gain further information about the role of Dni1p in the mating process, we observed cells carrying Dni1–GFP under the fluorescence microscope. We failed to observe any fluorescent signal in the cells during vegetative growth. When we treated h-cyr1Δsxa2Δ cells (which are responsive to pheromones in nitrogen-containing medium and hypersensitive to P factor; Stern and Nurse, 1997) carrying the GFP-fused Dni1 protein with the h+ pheromone P factor, we found that most Dni1p was localized to the tip of the shmoos (Fig. 2B), showing that intercellular contact was not required for the induction of Dni1p. When h90 cells were induced to mate, Dni1p was observed as a discrete patch at the area of cell–cell contact in the prezygotes (Fig. 2C, i), although it was occasionally possible to observe some fluorescent dots in the cell body (Fig. 2C, ii), or some uniform fluorescence spreading along the plasma membrane (see arrow in Fig. 2C, iii). Soon after cell fusion had started, the Dni1–GFP signal was observed as discrete dots that accumulated along the mating bridge of the zygotes (Fig. 2D). When cell fusion was complete, the protein was observed at the vacuoles, suggesting that it might have undergone some post-translational modification that rendered it unstable (Fig. 2E). These results were in agreement with the data obtained by Western blotting and suggested that Dni1p might have a specific role in the initial steps of mating.
To determine whether the Dni1 protein was expressed in both mating types, we performed mating assays between h+ or h- cells that carried the Dni1–GFP protein and untagged cells of the opposite mating type. We found that Dni1p localized to the cell–cell contact area when it was expressed in cells from either mating type (not shown).
Dni1p localization to the tip of shmoos requires Fus1p, actin and lipid rafts
We next wanted to study the requirements for Dni1p localization. First, we studied the relationship between dni1+ and other genes involved in cell fusion. We analysed the localization of Dni1p in cfr1Δ, cdc8-F41 and fus1Δ mutants under mating conditions. As shown in Fig. 3A, the localization of Dni1p in the control strain and in the cfr1Δ and the cdc8-F41 mutants was similar. In the fus1Δ mutant, we observed that Dni1p exhibited a polarized localization, although the protein was spread along the mating projection instead of being concentrated in a discrete patch at the shmoo tip. In contrast, Fus1p was properly localized in a dni1Δ mutant (Fig. 3B), suggesting that Dni1p could act downstream from Fus1p.
Since Fus1p is required for the organization of actin patches at the tip of the shmoos (Petersen et al., 1995; 1998b,c), we decided to investigate whether the localization of Dni1p depended on actin. To address this question, we treated cyr1Δsxa2Δ shmoos carrying the GFP-fused Dni1 protein with the actin-depolymerizing drug latrunculin A. After this treatment, Dni1p was polarized at the tip of the shmoos but did not concentrate in a patch (Fig. 3C), a result that showed that actin was required for the proper localization of Dni1p.
Since Dni1p is a transmembrane protein, we wished to gain information about the influence of membrane organization in its localization. To address this issue, we treated shmoos carrying GFP-tagged Bgs4p (an integral membrane protein) or Dni1p with filipin, an antibiotic that binds to sterols (Drabikowski et al., 1973) and that disrupts lipid rafts when the cells are incubated in its presence for long periods of time (Takeo, 1985; Wachtler et al., 2003). We observed that Bgs4p lost its localization to the projection of the shmoos when the lipid rafts had been disrupted by incubation in the presence of filipin for 1 h (Fig. 3D, left), in agreement with previous results (Wachtler et al., 2003). Similarly, Dni1p was not observed as a neat patch at the tip of the shmoos when the cells had undergone the same treatment (Fig. 3D, right). A similar result was obtained when the shmoos were incubated in the presence of the sterol-binding drug methyl-β-cyclodextrin for 1 h (not shown), showing that the integrity of lipid rafts was required for the proper localization of Dni1p.
dni1+ deletion leads to a defect in cell fusion
In order to investigate the role of Dni1p, we analysed the phenotype of a dni1Δ mutant and found that this strain showed no obvious phenotype during vegetative growth (data not shown). Next, we performed several experiments in order to know whether Dni1p played any role in mating and to pinpoint the step at which this protein might exert its function. We found that dni1Δ cells were able to produce both mating pheromones, and that shmoos were produced in response to P factor with the same efficiency and kinetics in the cyr1ΔsxaΔ control cells and in the cyr1ΔsxaΔdni1Δ mutant (data not shown). We next wondered if the dni1Δ cells were able to agglutinate in response to nitrogen deprivation, and we found that the agglutination index was the same for the h90dni1Δ mutant and for the h90 control strain, and that the Map4p agglutinin localized properly in the dni1Δ cells (not shown).
When we analysed the mating process in the h90 and the h90dni1Δ cells on solid EMM medium, we found that at the time at which the h90 strain had sporulated the dni1Δ mutant had produced zygotes with long mating bridges, but mature asci were scarce (see Fig. 4A). In quantitative analyses, we found that the efficiency of mating initiation (this parameter, measured as the number of zygotes plus asci with respect to the number of zygotes, asci, and cells, reflected the number of cells that were able to differentiate and establish a stable cell–cell contact) was similar in the h90 and the h90dni1Δ strains (27% and 26%, respectively, n = 500). However when we quantified the sporulation efficiency (this parameter, measured as the number of asci with respect to the number of asci plus zygotes, reflected the number of zygotes in which cell fusion, meiosis and spore development had proceeded properly), we found that in the wild-type (WT) strain 83% of the zygotes had produced asci after 36 h of incubation at 32°C while in the dni1Δ mutant strain this value was 6%.
We produced a dni1Δ/dni1Δ diploid strain and we found that it sporulated as efficiently as a control diploid strain (not shown). Since in S. pombe meiosis proceeds after karyogamy and zygotes give rise to asci immediately, this result suggested that the defect in sporulation exhibited by the h90dni1Δ mutant was a not a defect in meiosis or spore formation but the consequence of a defect in cell fusion. In order to confirm this hypothesis, we stained the cells with DAPI (for nuclear staining) and Calcofluor (for cell wall staining). As shown in Fig. 4B, at the time at which in the control strain the nuclei from both parental cells had fused, in the mutant strain nuclei were still apart from each other. Additionally, cell wall material could be observed between the cells. Concanavalin A staining confirmed that the cell wall was not digested at the cell–cell contact area (Fig. 4C). Finally, we introduced the GFP under the control of the map4+ promoter into the h90 or h90dni1Δ strains so that the fluorescent protein was only expressed in the h+ cells. We observed that in the WT strain the GFP was present throughout the cytoplasm of the zygotes (Fig. 4D), while in the dni1Δ mutant the fluorescent signal was only observed on one side of the zygote body of most zygotes (Fig. 4D). In the S. cerevisiae prm1Δ and fig1Δ mutants, which are defective in membrane fusion, it is possible to observe intercellular bubbles in which the fluorescence corresponding to the soluble GFP invades the cell body of one of the mating partners (Heiman and Walter, 2000; Aguilar et al., 2007). We observed this kind of structure (see arrow in Fig. 4D) in less than 10% of the dni1Δ zygotes. All these results showed that Dni1p was involved in the cell fusion process and suggested that the lack of fusion was due to a defect in cell wall digestion. Thus, the defect in sporulation observed in the mating mixtures involving the dni1Δ mutant was a consequence of a defect in cell fusion. This result would allow sporulation efficiency to be used as an indirect measurement of cell fusion efficiency in the experiments described below.
With a view to analysing whether the Dni1p function was mating type-associated, we produced h+ and h-dni1Δ haploid cells and performed unilateral and bilateral crosses. We found that in all crosses cells were able to initiate mating and produce zygotes with similar efficiency (Fig. 4E). Regarding the sporulation efficiency, this was 76% for the h- WT × h+ WT, 61% for the h- WT × h+dni1Δ, 43% for the h-dni1Δ x h+ WT and 10% for the h-dni1Δ × h+dni1Δ crosses (Fig. 4E), showing that the presence of Dni1p in one of the mating cells was sufficient to support cell fusion with reasonable efficiency.
This result allowed us to use a genetic approach to confirm the relationship between fus1+ and dni1+. We compared the sporulation efficiency in bilateral WT × WT, dni1Δ × dni1Δ and fus1Δ × fus1Δ crosses with that obtained in unilateral dni1Δ × WT, WT × fus1Δ and dni1Δ × fus1Δ crosses. If both genes acted at different points of the same pathway, the result of the unilateral dni1Δ × fus1Δ cross would be similar to that of the bilateral fus1Δ × fus1Δ cross. In contrast, if fus1+ and dni1+ were acting in parallel pathways, the result of the dni1Δ × fus1Δ cross would be similar to that of the unilateral dni1Δ × WT or WT × fus1Δ crosses. As shown in Fig. 4F, sporulation efficiency in the unilateral dni1Δ × fus1Δ cross was similar to that obtained in the bilateral fus1Δ × fus1Δ cross, strongly suggesting that dni1+ and fus1+ might act in the same pathway. As a control, the result of a cross involving fus1Δ and cfr1Δ, which act in different pathways (Cartagena-Lirola et al., 2006), was included.
Membrane fusion and organization, and cell wall remodelling are defective in dni1Δ mutants
All the above results showed that Dni1p was essential for the cell fusion step of mating and pointed to a role of this protein in cell wall digestion or remodelling. In order to gain information about this role, we analysed h90 and h90dni1Δ cells that had been induced to mate at 32°C, using transmission electron microscopy (Fig. 5). We observed that in the control strain some membranous structures accumulated at the cell–cell contact area when the cell wall between the parental cells was still present (Fig. 5A, i). At later stages, the cell wall could not be observed at the fusion area and the membranes from both cells were apposed (Fig. 5A, ii). At the final stage of cell fusion, no membrane or cell wall separated the parental cells (Fig. 5A, iii). In the dni1Δ mutant, several abnormalities were observed (Fig. 5B). An abnormal accumulation of membrane material apposed to the cell wall was present in some cells before cell wall digestion had started (Fig. 5B, i). In some cases, the cell wall was partially digested but abnormal membranous structures were observed between the parental cells (Fig. 5B, ii, iii, iv). It was sometimes possible to observe intercellular membrane bubbles that invaded one of the mating cells (Fig. 5B, iii, iv) or intracellular membrane bubbles that grew into one of the mating partners when the cell wall delimiting both cells had not been digested (Fig. 5B, v, vi). In some cases, the abnormal membranous structures seemed to contain cell wall material (Fig. 5B, ii, iii, iv, v). At later stages, it was possible to observe that cell wall material had been synthesized around the membrane bubbles (Fig. 5B, vii), and that some cytoplasmic blebs had been trapped by cell wall material (Fig. 5B, vii, viii, ix). These results revealed that in the absence of Dni1p the plasma membrane and cell wall were severely disorganized in the mating cells, and that Dni1p was required for membrane organization and cell wall remodelling during mating.
In order to determine whether Dni1p was a scaffold for the localization of enzymes required for the synthesis and/or degradation of the cell wall at the mating projection, we analysed the localization of the Bgs1p, Bgs3p and Bgs4p glucan synthases, and that of the Agn1p and Eng1p glucanases during mating in the h90 and the h90dni1Δ cells. We did not observe Eng1p in the zygotes from either strain, which is in agreement with the reported downregulation of the expression of this gene at the time of cell fusion (Mata et al., 2002). In contrast, Bgs1p, Bgs3p, Bgs4p and Agn1p were observed at the cell–cell contact area in both strains (Fig. S2 and results not shown). We also found that Cdc42p, a GTPase involved in polarity and cell wall synthesis, was localized to the cell–cell contact area in the WT and the dni1Δ mutant (Fig. S2 and results not shown). These results showed that Dni1p is not a general scaffold for the localization of enzymes required for cell wall remodelling.
Dni1p function is more relevant at temperatures above 25°C
During the development of these studies, we observed that the strength of the cell fusion defect in the dni1Δ cells depended on the temperature. To further study this, we quantified the sporulation efficiency in h90, h90dni1Δ and h90fus1Δ cells that had been induced to mate at different temperatures (20–37°C; see Fig. 6A). We found that in the WT strain sporulation was reduced at temperatures above 35°C, and that the fus1Δ strain exhibited a severe sporulation defect (less than 1% of the zygotes produced asci) at all the temperatures tested. In the dni1Δ strain, sporulation was efficient at 20°C, reduced at 25°C (50% with respect to the WT), and severely reduced at higher temperatures. Analysis of the sporulation process in dni1+/dni1+ and in dni1Δ/dni1Δ diploids confirmed that meiosis and spore development proceeded normally in the absence of Dni1p at all temperatures tested (see Fig. 6B). Finally, observation of zygotes in which the GFP was expressed under the control of the map4+ promoter confirmed that cell fusion was indeed more sensitive to temperature in the dni1Δ mutant than in the WT strain (not shown). These results showed that Dni1p was dispensable for cell fusion at 20°C. In order to determine whether Dni2p could undertake the function of Dni1p at low temperatures, h90, h90dni1Δ, h90dni2Δ and h90dni1Δdni2Δ strains were induced to mate at different temperatures. We found that at 32°C the dni1Δ, the dni2Δ and the dni1Δdni2Δ strains exhibited a similar defect in the efficiency of sporulation, and that at 20°C sporulation was as efficient in the three mutant strains as in the h90 control strain (Fig. 6C). The absence of intercellular diffusion of a soluble GFP confirmed that the defect in sporulation observed in the h90dni2Δ and h90dni1Δdni2Δ strains was due to a defect in cell fusion (Fig. 6D). These results showed that Dni1p and Dni2p cannot substitute each other and suggested that these proteins act together in a fine-tuned mechanism directed to ensuring proper membrane organization and cell wall remodelling during cell fusion at higher temperatures, at which membranes are more fluid and biochemical reactions proceed faster.
In S. pombe, calcium is not accumulated in response to pheromones and the external calcium concentration does not affect cell survival upon sexual differentiation
Since Dni1p shares significant similarity with the S. cerevisiae Fig1 protein, which is involved in calcium uptake and cell fusion (Erdman et al., 1998; Muller et al., 2003; Aguilar et al., 2007), we wondered whether Dni1p was involved in calcium uptake during mating. To answer this question calcium accumulation analyses were performed by incubating cyr1Δsxa2Δ and cyr1Δsxa2Δdni1Δ cells with 45Ca++ in the presence or absence of the h+ pheromone P factor (see Experimental procedures). As a control, we used an ehs1Δ strain (Carnero et al., 2000), which lacks a Ca++-permeable channel (Carnero et al., 2000; Tasaka et al., 2000). The experiment was performed by incubating the cells in medium with or without calcium for different times at 32°C. As shown in Fig. 7A, after 4 h of incubation the cells accumulated more 45Ca++ in the calcium-devoid medium than in the calcium-containing medium, the opposite result from that obtained for different strains in S. cerevisiae (Iida et al., 1990). As expected, the ehs1Δ strain did not accumulate calcium in any condition tested (Fig. 7A). In the WT strain, cells accumulated more calcium when they were treated with pheromone than when they were not, although the difference between both conditions was not as dramatic as that described for S. cerevisiae (Iida et al., 1990). Surprisingly, the behaviour of the dni1Δ strain was similar to that of the control strain, it being able to accumulate calcium in the presence and the absence of P factor (Fig. 7A). Similar results were obtained after 1 h and after 6 h of incubation (not shown).
It has been described that S. cerevisiae cells die when they differentiate into shmoos either in a calcium-deficient medium or in cells lacking the High-Affinity Ca++ influx System (HACS) channel Mid1p (Iida et al., 1990; 1994). In order to determine the effect of the external calcium concentration on the viability of S. pombe cells treated with pheromone, we exposed cyr1Δsxa2Δ and cyr1Δsxa2Δdni1Δ cells to P factor in EMM, EMM-Ca, and EMM-Ca supplemented with 10 mM BAPTA (a calcium-binding agent) or with 100 mM calcium. Samples were taken every hour over 6 h and used for microscopic observation and for viability determination by staining the cells with methylene blue and by quantifying the colony-forming units (cfu). We found that cells from both strains produced shmoos with the same efficiency and kinetics, and that more than 90% of these shmoos were viable in all conditions tested (not shown). We also used the methylene blue-staining method to determine zygote lysis in h90 and h90dni1Δ strains that had been induced to mate in EMM, EMM-Ca, and EMM-Ca supplemented with 10 mM EGTA (a calcium-chelating agent). We found that in both strains lysis was about 10% in all conditions tested. All these results indicated that external calcium was not required for maintaining the viability of S. pombe cells in response to pheromone.
Finally, we analysed the effect of the external calcium concentration on cell fusion. To do so, the sporulation efficiency of an h90dni1Δ strain was compared with that of the h90 control strain. We found that calcium depletion did not reduce the sporulation efficiency in either the h90 or the h90dni1Δ strains. Additionally, the sporulation of dni1+/dni1+ and dni1Δ/dni1Δ diploid strains was efficient on EMM, EMM-Ca, and EMM-Ca plates supplemented with 10 mM EGTA or 100 mM calcium (not shown). These results show that the external calcium concentration does not have a significant influence in the cell fusion process in S. pombe. Surprisingly, the addition of 100 mM calcium, but not of 100 mM magnesium, partially rescued the defect in sporulation of the dni1Δ mutants (Fig. 7B). The same phenomenon has been observed for the fig1Δ mutant. It is possible that such high calcium concentrations could induce some cell response that had an indirect effect in cell fusion.
In this work, we have investigated the function of Dni1p. dni1+ expression is induced in response to nitrogen starvation (Mata et al., 2002), suggesting that Dni1p could play some role in mating. In agreement with this notion we found that this protein accumulated during the initial steps of mating; Dni1p localized to the tip of the shmoos and the mating bridge, and dni1Δ cells were defective in cell fusion. We performed several experiments to determine the nature of the mating defect in this strain. It has been described that a functional cytoskeleton is required for cell fusion in S. pombe (Petersen et al., 1998a,b; Kurahashi et al., 2002). We found a genetic interaction between dni1+ and fus1+, a gene coding for a formin homologue that is required for the organization of actin patches during mating (Petersen et al., 1998b). Immunolocalization analyses showed that microtubules and actin patches were normally distributed in dni1Δ zygotes (not shown), and in vivo observation of a dni1Δ mutant carrying a GFP-tagged Crn1 protein (Fig. S3, Coronin, an actin patch-associated protein; Pelham and Chang, 2001) confirmed the latter result. Additionally, Fus1p was properly localized in the dni1Δ mutant (Fig. 3). All these results indicated that Dni1p does not play a role in organizing the cytoskeleton during mating. Dni1p was not properly localized in a fus1Δ mutant (Fig. 3). Given the general role of formins and the actin cytoskeleton in delivering the majority of membrane proteins via appropriately polarized vesicle trafficking, this result could be interpreted as a dependence of Dni1p localization on actin organization and cell polarity. In fact, actin depolymerization also resulted in Dni1p delocalization (Fig. 3), in support of this idea. A multicopy plasmid carrying dni1+ did not alleviate the mating defect of a fus1-B20 strain carrying a point mutation in fus1+ (Petersen et al., 1995; not shown), which suggested either that Dni1p could act downstream of Fus1p or that in order to perform its activity Dni1p requires other protein(s) that depend on Fus1p. We found that actin was required for Dni1p localization but, surprisingly, a mutation in tropomyosin (cdc8-F41 strain; Kurahashi et al., 2002) did not affect that localization. It has been shown that Cdc8p is indispensable for cell fusion during conjugation in fission yeast and it has been speculated that tropomyosin might organize a small F-actin-containing organelle at the cell fusion site (Kurahashi et al., 2002). It is possible that the cdc8-F41 point mutation might only affect the function of some of proteins localized in that organelle or that this organelle does not contain Dni1p.
Electron microscopy shed light on the function of Dni1p. In the absence of this protein, there was a severe disorganization of the plasma membrane and the cell wall at the cell–cell contact area. Some cells exhibited abnormal membranous structures at the fusion region. We also observed intercellular bubbles and cytoplasmic blebs embedded in the cell wall. This phenotype was not a non-specific consequence of the lack of cell fusion, since in a cfr1Δ mutant electron microscopy revealed that the cell wall remained undigested between the parental cells, but intercellular bubbles and blebs were not observed (Cartagena-Lirola et al., 2006). This heterogeneous phenotype could be explained if Dni1p were a regulator of several proteins required for different processes, or if the function of Dni1p were to ensure the co-ordination between different steps of cell fusion and some of these steps could occasionally proceed in the absence of this protein. The fact that Dni1p function was found to be dispensable at low temperatures, at which biological processes proceed slowly, points in the latter direction. Additionally, we found that the Bgs1p, Bgs3p, Bgs4p glucan synthases and the Agn1p glucanase localized to the mating projections in a dni1Δ mutant (Fig. S2), suggesting that although Dni1p could be required for the localization of some protein involved in cell wall remodelling, it is not a general scaffold for the localization of the enzymes required for this process.
Dni1p is similar to the S. cerevisiae calcium uptake regulator Fig1p. We therefore investigated whether Dni1p and Fig1p were homologues, and we found that there exist a number of similarities and differences between these proteins and their functionalities. When we analysed the relationship between mating, calcium uptake and Dni1p we found that calcium accumulation and cell survival, upon exposure of cells to mating pheromones in the absence of calcium, were similar in the WT and dni1Δ strains. This is in agreement with the fact that Fig1p is a component of the LACS system (Muller et al., 2003), which mediates calcium uptake in the presence of nitrogen, a condition that inhibits mating in S. pombe (Yamamoto et al., 1997; Davey, 1998; Nielsen, 2004). These results suggested that the main function of Dni1p in cell fusion is not to regulate a calcium influx in response to pheromones and that Fig1p and Dni1p are not functional homologues. However, whereas it has been described that in S. cerevisiae a strong intracellular calcium accumulation takes place in response to pheromones, and that this calcium accumulation is essential for cell survival upon shmoo formation (Iida et al., 1990; 1994), we failed to detect a strong accumulation of calcium or reduced viability in a calcium-devoid medium in response to pheromones in a S. pombe WT strain (Fig. 7). Additionally, we found that the absence of external calcium did not influence cell fusion during mating in a WT S. pombe strain (Fig. 7B). It therefore seems that the effect of the external calcium concentration has a different impact on the sexual development on both organisms, a result that would account for the different functionality of Fig1p and Dni1p in regard to calcium uptake/accumulation.
Although Fig1p function was initially associated to the regulation of a calcium-influx system during mating (Muller et al., 2003), it was later described that this protein is also required for the co-ordination between cell wall remodelling and membrane fusion during mating (Aguilar et al., 2007). Additionally, in C. albicans, Fig1p is involved in calcium influx and polarized growth (Brand et al., 2007; 2009). It is not known whether the calcium-influx activity of Fig1p affects cell polarization in S. cerevisiae and in C. albicans, or whether these are separate functions of the proteins. Additionally, it is still unknown how other proteins work with Fig1p to accomplish certain aspects of polarization, calcium influx and cell–cell fusion, and how these processes might be interrelated. Thus, it is possible that that Dni1p and Fig1p might function in a related way and that the co-ordination between cell wall remodelling and membrane fusion, the function that we have proposed for Dni1p and that has also been suggested for Fig1p (Aguilar et al., 2007), is the function most conserved along the evolution of these proteins. Accordingly, either Dni1p represents an out-group of this family of proteins or it is a true homologue of this family but its particular functionality reflects the differences between the budding and fission yeast lineages and their requirements for mating. The evolutionary divergence of these proteins and their partners probably accounts for the fact that Fig1p, expressed under the control of the dni1+ promoter and terminator sequences in S. pombe, localized along the shmoo projections but was unable to complement the mating defect of a dni1Δ strain (results not shown). Proteins similar to Dni1p and Fig1p are present in the genomes of other yeasts and fungi. Determining the function of these proteins should help to understand the main role of this family of proteins.
In S. cerevisiae, genetic and cytological analyses of mutants defective in mating have led to the establishment of several parallel subpathways that affect different steps of cell–cell fusion (cell signalling, polarity, cell wall remodelling and plasma membrane fusion; Gammie et al., 1998; Ydenberg and Rose, 2008). In S. pombe, only two pathways required for cell fusion during mating have been reported; one of them includes the formin homologue fus1+ and is involved in actin organization and polarity (Petersen et al., 1995; 1998c). The other pathway is defined by the Golgi protein Cfr1p, which might be involved in the delivery of a cell wall-degrading enzyme (Cartagena-Lirola et al., 2006). dni1Δ mutants exhibit aberrant cell wall and membrane structures at the cell–cell contact area. Some of these structures are reminiscent of those observed in S. cerevisiae prm1Δ, fig1Δ, prm1Δfig1Δ and prm1Δkex2Δ mutants (Heiman and Walter, 2000; Aguilar et al., 2007; Heiman et al., 2007), suggesting that S. pombe Dni1p might share a function with Fig1p, Prm1p and/or a Kex2-proccessed protein in cell wall remodelling and plasma membrane fusion, defining a new cell fusion subpathway in the fission yeast. In S. pombe, the protein encoded by the SPBC4.01 ORF exhibits sequence similarity to Dni1p and Fig1p (Fig. S1). We analysed whether this protein (Dni2p) might have a redundant role with Dni1p at low temperatures and we found that the single dni2Δ and double dni1Δdni2Δ mutants have a temperature-sensitive mating defect equivalent to that of dni1Δ mutants. Importantly, the phenotype of the double mutant was no more severe than that of either single mutant. This rules out the hypothesis that the Dni1 and Dni2 proteins functionally substitute for each other, and suggests rather that they have separate essential roles in the same fusion subpathway.
Tetraspan proteins participate in different cellular processes and are sometimes associated with some membrane subdomains (Gonzalez-Mariscal et al., 2003; Stipp et al., 2003; Tarrant et al., 2003; Yunta and Lazo, 2003; Hemler, 2005). Dni1p is a tetraspan similar to the Fig1 and the Sur7 families of claudin-related proteins (Fig. 1; Muller et al., 2003; Walther et al., 2006; Zhang et al., 2006; Alvarez et al., 2008; Grossmann et al., 2008). Claudins are the most important components of the tight junctions, where they establish a barrier that controls the flow of molecules in the intercellular space and block the movement of integral membrane proteins. In C. albicans, deletion of SUR7 leads to several phenotypes, including defective endocytosis and an abnormal synthesis of cell wall material that grows into the cytoplasm (Alvarez et al., 2008). Endocytosis of the Map3 pheromone receptor and uptake of FM4-64 seems to take place normally in a dni1Δ mutant (Fig. S4), but an abnormal growth of the cell wall into the cytoplasm was observed in this strain during mating, suggesting that CaSur7p and Dni1p could share some function. Dni1p was observed as a discrete patch, whose integrity required lipid rafts, at the cell fusion area (Fig. 3). Thus, Dni1p could define a specialized membrane microdomain that would ensure an adequate spatiotemporal regulation of the membrane organization, cell wall remodelling and cell–cell communication during cell fusion. In S. cerevisiae, it has been shown that the composition and distribution of membrane subdomains play a relevant role in mating (Bagnat and Simons, 2002; Proszynski et al., 2006; Jin et al., 2008). The situation is probably similar in S. pombe. Finally, it is noteworthy that tetraspanins (a family of tetraspan proteins that cluster in characteristic membrane subdomains; Yunta and Lazo, 2003; Hemler, 2005) play an essential role in sperm-egg fertilization in mammals (for a review see Sutovsky, 2009) and that tetraspan proteins (Fig1p, Prm1p and Dni1p) are required for correct membrane reorganization during mating in yeasts (this work; Heiman and Walter, 2000; Jin et al., 2004; Aguilar et al., 2007; Heiman et al., 2007). A detailed analysis of the role of Dni2p and Prm1p in S. pombe, and their relationship with Dni1p, should help to provide information about the relevance of fungal tetraspan proteins in mating.
Strains and growth conditions
All general growth conditions and yeast manipulations have been described previously (Moreno et al., 1991; http://www.biotwiki.org/bin/view/Pombe/NurseLabManual). S. pombe strains are derivatives of the 972 h- and 975 h+ WT and were grown in YES or EMM. EMM-Ca was prepared by replacing calcium pantothenate by sodium pantothenate. EMM-N was EMM without ammonium chloride. BAPTA and EGTA were from SIGMA.
Pheromone production was assessed as described previously (Egel et al., 1994). Shmoo formation was induced in cyr1Δsxa2Δ strains by adding synthetically synthesized P factor (Isogen; final concentration of 1.5 μg ml−1). cyr1Δsxa2Δ cells are sensitive to low concentrations of pheromones in nitrogen-containing media because they lack the serine carboxypeptidase sxa2+, which degrades P factor, and the adenylate cyclase cyr1+ (Stern and Nurse, 1997). Shmoos were treated with 100 μM, final concentration, of Latrunculin A (SIGMA) for 10 min. Agglutination tests were carried out in EMM-N as described previously (Sharifmoghadam and Valdivieso, 2008). Cells were induced to mate on EMM or EMM-Ca plates at different temperatures. Mating efficiency represented the number of zygotes plus asci with respect to the total cell number (zygotes plus asci plus vegetative cells), and sporulation efficiency was the number of asci with respect to the number of asci plus zygotes, as described in Arellano et al. (2000).
Molecular and genetic manipulations
The dni1+ ORF and 1 kb DNA fragments corresponding to the 5′ and 3′ non-coding regions were amplified by PCR, cloned into the KS+ vector (Stratagene), and sequenced. A dni1Δ deletion cassette, in which the complete dni1+ ORF had been substituted by the KanMX6 gene that confers resistance to geneticin (Bähler et al., 1998), was used to transform the strains of interest. Correct integration was always assessed by PCR. Site-directed mutagenesis was used to introduce a NotI restriction site immediately upstream from the stop codon. The GFP was cloned as NotI/NotI DNA fragment. The GFP-tagged dni1+ allele, under the control of its own promoter, was integrated at the leu1+locus. A soluble GFP was cloned, as an NdeI/NotI DNA fragment, into an integrative plasmid that carried the map4+ promoter and terminator sequences and that had been digested with those restriction enzymes. The S. cerevisiae FIG1 gene was PCR-amplified, sequenced and cloned as a SmaI/NotI DNA fragment into an integrative plasmid, which carried the dni1+ promoter and terminator sequences and which had been digested with those restriction enzymes. The GFP was cloned at the NotI site of the latter plasmid. A dni2::ura4+ deletion cassette was constructed by cloning 1 kb DNA fragments corresponding to the dni2+ 5′ and 3′ non-coding regions (which had been PCR-amplified) upstream and downstream of the ura4+ gene, which was cloned in a KS+ vector.
Induction of mating in liquid medium and Western blotting was performed as described previously (Sharifmoghadam and Valdivieso, 2008). In order to detect the GFP-tagged Dni1 protein, cells were broken in 50 mM Tris HCl, pH 7.5, 300 mM NaCl, 50 mM EDTA supplemented with protease inhibitors (1 mM PMSF; 1 μg ml−1 Aprotinin, Leupeptin and Pepstatin), and cell debris was eliminated by centrifuging at 1600 r.p.m. for 5 min at 4°C. Then, cell extracts were concentration-equalled (total protein was estimated using the Bio-Rad protein assay kit; Bradford method), incubated at 0°C in the presence of 1.6 M urea for 20 min and centrifuged at 13000 r.p.m. for 30 min at 4°C. Samples were denatured in Laemmli sample buffer (50 mM HCl-Tris, pH 6.8; 1% SDS; 143 mM β-mercaptoethanol; 10% glycerol) at 42°C for 5 min. In order to identify unspecific bands and bands corresponding to cleaved GFP cell extracts from a strain that carried a soluble GFP under the control of the map4+ promoter were analysed in parallel. Anti-GFP and anti-Cdc2 were used at 1:1000 and 1:4000 dilutions respectively.
Calcium accumulation analysis
Calcium uptake and accumulation was measured as previously described (Iida et al., 1994; Paidhungat and Garrett, 1997). Basically, cyr1Δsxa2Δ cells were incubated in EMM or in EMM-Ca for 4 h at 30°C. Then, 3.8 μl of 45Ca++ (CES3, Amersham; 2.2 μCi μl−1) was added, the cells were treated or not with factor P as described above, and incubated for different times at 32°C. The reaction was stopped by mixing 0.2 ml of samples from the cultures with 2 ml of cold STOP buffer (100 mM Tris-HCl, pH 6.8/20 mM CaCl2). Samples were filtered and radioactivity was measured using a liquid scintillation counter. Non-radioactive cultures were performed in parallel and were used to estimate the cell number in each sample.
A Leica DM RXA microscope equipped with a Photometrics Sensys CCD camera using the Qfish 2.3 program was used to perform optic microscopy analyses. Calcofluor (BLANKOPHOR, Bayer) and DAPI (SIGMA) staining was performed as described (Arellano et al., 2000). For FITC-conjugated concanavalin A (SIGMA) staining, cells were washed twice with 1 ml of phosphate-buffered saline (PBS) and suspended in 50 μl of PBS. FITC–concanavalin A (5 μl of a stock solution at 1 mg ml−1) was added and the cells were incubated at room temperature for 15 min in the dark. After incubation, the cells were washed three times with 500 μl of PBS, re-suspended in 50 μl of the same buffer and observed under a fluorescence microscope. Actin staining with rhodamine-phalloidin was performed as in Marks and Hyams (1985), and anti-tubulin immunofluorescence was performed by fixing the cells in methanol as described in Hagan and Hyams (1988) with an anti-Tat1 antibody. Cell lysis was assessed by re-suspending the cells in 0.3% methylene blue and immediately observing them under the bright-field microscope. FM4-64 (BIOTIUM) staining was performed by adding this dye at a final concentration of 4 μM to the samples, incubating them for different times and washing the cells with cold EMM. Filipin (SIGMA) staining and treatment of cells was performed as described (Wachtler et al., 2003). Methyl-β-cyclodextrins (SIGMA) were used at 200 mM. For transmission electron microscopy, cells were fixed with 2% EM grade glutaraldehyde (GA; Electron Microscopy Science) in 50 mM phosphate buffer, pH 7.2, 150 mM NaCl (PBS) for 2 h at 4°C, post-fixed with 1.2% potassium permanganate overnight at 4°C and embedded in Quetol 653, as described (Konomi et al., 2003). Ultrathin sections were stained in 4% uranyl acetate and 0.4% lead citrate, and viewed under a TEM H-800 (Hitachi) operating at 125 keV.
We thank B. Santos and A. Duran for a critical reading of the manuscript and R. Egel, S. Erdman, O. Nielsen, P. Pérez, J.C. Ribas, Y. Sánchez, C. Shimoda, C.R. Vázquez de Aldana, M. Yamamoto and the Yeast Genetic Resource Center (Japan) for plasmids and strains. We are indebted to N. Skinner for language revision, to Y. Sánchez for communicating unpublished results and to P. Pérez for the anti-Tat1 antibody. This work has been supported by Grant BFU2007-61866 from the CICYT and Grants SA128/04, SA104A/07 and GR231 from the Junta de Castilla y León, Spain. J.Á.C.-R. and R.M.-G. were supported by fellowship from the MEC, Spain, and M.R.S. was supported by a fellowship from the Government of Iran.