Identification and characterization of LFD1, a novel protein involved in membrane merger during cell fusion in Neurospora crassa

Authors


Summary

Despite its essential role in development, molecular mechanisms of membrane merger during cell–cell fusion in most eukaryotic organisms remain elusive. In the filamentous fungus Neurospora crassa, cell fusion occurs during asexual spore germination, where genetically identical germlings show chemotropic interactions and cell–cell fusion. Fusion of germlings and hyphae is required for the formation of the interconnected mycelial network characteristic of filamentous fungi. Previously, a multipass membrane protein, PRM1, was characterized and acts at the step of bilayer fusion in N. crassa. Here we describe the identification and characterization of lfd-1, encoding a single pass transmembrane protein, which is also involved in membrane merger. lfd-1 was identified by a targeted analysis of a transcriptional profile of a transcription factor mutant (Δpp-1) defective in germling fusion. The Δlfd-1 mutant shows a similar, but less severe, membrane merger defect as a ΔPrm1 mutant. By genetic analyses, we show that LFD1 and PRM1 act independently, but share a redundant function. The cell fusion frequency of both Δlfd-1 and ΔPrm1 mutants was sensitive to extracellular calcium concentration and was associated with an increase in cell lysis, which was suppressed by a calcium-dependent mechanism involving a homologue to synaptotagmin.

Introduction

Cell–cell fusion is an essential process in most multicellular organisms and occurs during both sexual and asexual development. Fertilization in sexually reproducing organisms requires a cell–cell fusion event between gametes. Cell fusion is also a common developmental process in somatic tissues in animals and is required to form the multinucleate syncytial cells found in muscle and multinucleate osteoclasts formed by macrophage fusion (Aguilar et al., 2013). Somatic cell fusion is also important for the development of the interconnected mycelial network characteristic of filamentous fungi (Fleissner et al., 2008).

Cell fusion to form syncytia requires membrane merger. The fusion of membranes requires lipid flow between tightly apposed membrane bilayers (Sollner, 2004). However, lipid bilayers will not spontaneously fuse, even when closely adhered. Biological membrane fusion is facilitated by proteins termed fusogens, which destabilize lipid bilayers, thereby assisting membrane merger. In intracellular vesicle trafficking, SNARE proteins facilitate membrane merger by forming coiled-coil interactions that distort and destabilize membrane lipid packing (Jahn et al., 2003; Sollner, 2004). Although the function of SNARE proteins and molecular mechanisms of intracellular vesicle fusion are well characterized, only a few fusogens have been identified that promote membrane merger during cell–cell fusion in eukaryotic systems. For example, in Caenorabditis elegans, two proteins, EFF-1 and AFF-1 are necessary and sufficient to fuse cells. EFF-1 is required for epithelial cell membrane fusion (Mohler et al., 2002) and AFF-1 is required for anchor cell fusion (Sapir et al., 2007). Both proteins are type I membrane glycoproteins belonging to a family of homotypic cell–cell fusogens (FF) (Podbilewicz et al., 2006; Sapir et al., 2007). The substitution of fusogens from enveloped viruses with C. elegans FF proteins compensated for viral fusogen function (Avinoam et al., 2011). However, homologues of FF proteins are only present in nematodes, arthropods, ctenophores, chordates and protists, and have not been found in other eukaryotic organisms (Avinoam et al., 2011).

The mating pathway of Saccharomyces cerevisiae has been extensively characterized and culminates in cell and nuclear fusion between cells of opposite mating type (White and Rose, 2001). Although mechanisms mediating cell communication, signal transduction and polarization during mating in S. cerevisiae are well characterized, only a few proteins involved in membrane merger have been identified. These include Fus1p, Fus2p, Fig1p and Prm1p. Prm1p and Fig1p have been implicated in formation of the fusion pore, while Fus1p and Fus2p localize to the expanding fusion pore (Elion et al., 1995; Erdman et al., 1998; Heiman and Walter, 2000; Proszynski et al., 2006; Aguilar et al., 2007; Heiman et al., 2007; Paterson et al., 2008). A significant number of mating pairs between strains bearing deletions of PRM1 or FIG1 show tightly apposed plasma membranes (Heiman and Walter, 2000; Aguilar et al., 2007). In addition, cell lysis of prm1Δ or fig1Δ mating pairs is common and is affected by extracellular calcium conditions (Jin et al., 2004; Aguilar et al., 2007).

The filamentous fungus Neurospora crassa consists on a syncytial network of highly polarized and interconnected multinucleate hyphae (Roper et al., 2013). Fusion between cells of opposite mating type occurs during fertilization, but cell fusion also occurs between genetically identical cells during vegetative growth. For example, when genetically identical N. crassa asexual spores germinate in proximity to each other, they undergo chemotropic growth, ultimately resulting in cell fusion (Fleissner et al., 2008; 2009b; Read et al., 2009). Cell fusion during vegetative growth facilitates the formation of the interconnected hyphal network that is characteristic of filamentous fungi and is important for nutrient and organelle translocation throughout the colony (Simonin et al., 2012; Roper et al., 2013).

The process of cell fusion in N. crassa follows three consecutive steps. The first is recognition and chemotropic growth between cells, the second step is adhesion and cell wall breakdown, and the last step is plasma membrane merger, fusion pore formation and cytoplasmic mixing (Hickey et al., 2002). In N. crassa, a PRM1 orthologue has been shown to play a role during both vegetative and mating cell fusion (Fleissner et al., 2009a). Somatic cell fusion frequency in the N. crassa ΔPrm1 mutant is reduced by ∼ 50%, remarkably comparable to the ∼ 50% reduction in mating cell fusion observed in the S. cerevisiae prm1Δ mutant (Heiman and Walter, 2000). TEM analyses of the N. crassa ΔPrm1 mutant during germling fusion showed tightly apposed plasma membranes in the contact zone, a phenotype never observed for cell fusion between wild type cells (Fleissner et al., 2009a). However, unlike the S. cerevisiae prm1Δ mutant, the N. crassa ΔPrm1 strain shows post-fertilization defects, suggesting that additional fusion events are regulated by PRM1 during sexual development (Fleissner et al., 2009a).

The observation that 50% of N. crassa germlings still undergo somatic cell fusion in the ΔPrm1 mutant suggests that additional components required for membrane merger remained to be discovered. We therefore mined transcriptional profiles of wild type versus a N. crassa strain carrying a deletion of a transcription factor, PP1, which regulates expression of Prm1 in germlings (Leeder et al., 2013). Notably, Δpp-1 mutants are completely defective for chemotropic interactions and germling fusion. Using this approach, we screened 33 strains carrying deletions of candidate genes for cell fusion defects. We identified a novel gene encoding a single pass transmembrane protein, late fusion defect-1 (lfd-1), which is involved in membrane fusion events during both somatic and mating cell fusion in N. crassa. Our data indicate that LFD1 is part of the cell fusion machinery and acts independently of PRM1. We determined that extracellular calcium is important for prevention of cell lysis, which occurs during unproductive fusion events in the absence of PRM1 or LFD1. Furthermore, we provide the first report of a possible role for a synaptotagmin homologue (SYT1) in membrane repair in filamentous fungi, and which is required to alleviate calcium-dependent cell lysis that occurs during defective cell fusion in ΔPrm1 and Δlfd-1 germlings.

Results

Identification of lfd-1

To identify candidate genes that play a role in late events in cell fusion, we performed a targeted analysis of RNA-seq data from wild type N. crassa (FGSC 2489) versus a strain carrying a deletion of a transcription factor Δpp-1, which is required for expression of Prm1 (Leeder et al., 2013). We specifically assessed expression data from germlings at a time point when chemotropic interactions and cell fusion in wild type were most prevalent (5 h post germination). Expression of Prm1 (NCU09337) was fairly low in wild type germlings (∼ 45 RPKM) at this time point, but was negligible in the Δpp-1 mutant (Fig. 1). From these analyses, we identified 33 candidate genes that showed a similar expression pattern as Prm1, with less than 5% of the expression level of wild type germlings in the Δpp-1 mutant.

Figure 1.

Transcriptomic profile of genes with similar expression patterns to Prm1 (NCU09337) in wild type germlings (black) undergoing chemotropic interactions and cell fusion versus Δpp-1 germlings (grey), that lack chemotropic interactions and cell fusion (Leeder et al., 2013). RPKM = reads per kilobase per million reads (Normalized counts).

The germling fusion phenotype of deletion strains for the 33 candidate genes was assessed using the plasma membrane dye FM4–64. Of the 33 candidates, strains carrying deletions of six genes (NCU09307, NCU05798, NCU05629, NCU05681, NCU03863 and NCU06912) showed defects in germling cell fusion. Five of these genes encode hypothetical proteins, two of which have predicted transmembrane (TM) domains (NCU09307 and NCU05798). The sixth gene, NCU03863, encodes a predicted protein with 4 TM domains and is the homologue of S. cerevisiae RBS1, which encodes an integral membrane protein believed to function in sphingoid long chain base transport (Kihara and Igarashi, 2002). One of the mutants, ΔNCU09307, showed membrane invaginations (Fig. 2A), a phenotype similar to ΔPrm1 mutants (Fleissner et al., 2009a), suggesting that ΔNCU09307 might be defective for membrane merger. The other five candidate deletion strains showed possible cell lysis during cell fusion (see below), but membrane invaginations were not observed.

Figure 2.

The ΔNCU09307 strain (Δlfd-1) shows membrane fusion defects.

A. FM4–64 stained germling fusion pairs showing normal membrane fusion (left) versus germling fusion pairs showing defective membrane fusion (right). Arrow points to a membrane invagination during defective membrane merger. Scale bars = 5μm.

B. Fusion between germlings expressing different cytoplasmic markers, used to confirm membrane fusion defects. Wild type (left) show complete mixing of GFP and dsRED, while ΔPrm1 (centre) and Δlfd-1 (right) germlings show blockage of cell fusion (arrows). Scale bars = 5 μm.

C. Membrane fusion frequency of wild type versus Δlfd-1 germlings, as compared with wild type + Δlfd-1 germlings or wild type + Δlfd-1 (his3::ccg1-dsRed-lfd-1) germlings. Asterisks indicate significant difference with respect to wild type (t-test P < 0.05, n = 3). Graph bars = SD.

To obtain quantitative data for cell fusion events, we constructed duplicate strains for each of the six candidate deletion strains, where one deletion strain carried a cytoplasmic GFP marker and another strain in the same deletion background carried a cytoplasmic dsRed (or mCherry) marker. For fusion assays, conidia from the same deletion strain, but carrying different markers (GFP or dsRed/mCherry) were mixed in an equal ratio. Five hours post germination, the fusion phenotype between green and red fluorescent germling pairs was analysed. If germlings had undergone fusion, red and green fluorescence was detected in both cells of the fusion pair. As shown in Fig. 2B, mixing of red and green cytoplasm was always observed in wild type germling pairs following chemotropic interactions and cell adhesion. In contrast, if the pair remained unfused, one germling remained red and the other remained green. Membrane invaginations were apparent in control pairings of ΔPrm1 germlings, as well as in strains carrying a deletion for NCU09307 (Fig. 2B, arrows).

To obtain quantitative data on fusion frequency, we evaluated three replicates of 50 pairs of adhered germlings for fusion in wild type and the ΔNCU09307 strain. While 100% of the adhered wild-type pairs (A9 + A10) fused and showed mixed green and red signals, fusion between adhered ΔNCU09307 germlings (JPG7 + JPG8) was reduced to 77% (Fig. 2C). The percentage of fused germlings was not significantly increased by extended incubation times (data not shown). Thus, similar to PRM1 (Fleissner et al., 2009a), LFD1 contributes to, but is not essential for germling fusion in N. crassa. Strains carrying deletions for the other five candidate genes (NCU05798, NCU05629, NCU05681, NCU03863 and NCU06912) did not show defects in cytoplasmic mixing. These results showed that NCU09307 was involved in cell fusion/membrane merger; NCU09307 was named lfd-1, for late fusion defect-1.

Similar to Prm1, but unusual for most N. crassa genes (Galagan et al., 2003), lfd-1 lacks introns. lfd-1 is predicted to encode a protein of 1095 amino acids (aa) and contains one transmembrane domain (aa 638 to 660). Prediction algorithms TMHMM, TMpred and PSORT (Hofmann and Stoffel, 1993; Krogh et al., 2001; Horton et al., 2007) all support placement of the C terminus of LFD1 on the exterior of the cell (Fig. 3A). Twelve of seventeen predicted N-glycosylation sites (http://www.cbs.dtu.dk/services/NetNGlyc/) occur in the predicted extracellular C-terminus of LFD1. LFD1 lacks a signal peptide, but a dileucine motif was identified at position 297 (… DNEDELLT …). Dileucine motifs can function as an endocytosis internalization signal for membrane proteins and receptors and are involved in the recycling of plasma membrane proteins (Pandey, 2010). Additional protein domains in LFD1 were not identified. lfd-1 is conserved (from 29–99% identity) among filamentous ascomycete fungi (subphylum Pezizomycotina), but obvious homologues to lfd-1 were not identified in the Saccharomycotina (Saccharomyces) or Taphrinomycotina (Schizosaccharomyces) subclades.

Figure 3.

Schematic representation of LFD1 and Δlfd-1 phenotype.

A. Schematic representation of LFD1 showing the dileucine motif and predicted transmembrane domain (TM). Twelve of seventeen predicted N-glycosylation sites occur in the predicted extracellular region of LFD1.

B. Macroscopic phenotype of Δlfd-1 as compared with wild type and Δlfd-1; ΔPrm1 strains.

C. Radial growth (left) and aerial hyphae extension (right) of wild type, ΔPrm1, Δlfd-1 and Δlfd-1; ΔPrm1 mutants. Asterisks indicate a significant difference with respect to wild type (t-test P < 0.05, n = 3). Graph bars = SD.

D. Dynamics of MAK2–GFP oscillation to CAT tips of Δlfd-1 germlings (JPG19) during chemotropic interactions are indistinguishable from oscillation of MAK2–GFP in wild type germlings (AF-M512) (Fleissner et al., 2009b). Ten separate events were scored. Arrows indicate MAK2–GFP localization. Scale bar = 5 μm.

Macroscopically, the Δlfd-1 mutant was indistinguishable from wild type (Fig. 3B). The average growth rate of the Δlfd-1 mutant [8.3 ± 0.2 cm/day (± SD)] was comparable to its isogenic wild-type parent [8.6 ± 0.6 cm/day (± SD); Fig. 3C]. We also assessed aerial hyphae extension in the Δlfd-1 mutant, as this trait is often reduced in hyphal fusion mutants of N. crassa (Fu et al., 2011). However, no significant difference (t-test, P > 0.05) from wild type in aerial hyphae extension was observed for either the ΔPrm1 or Δlfd-1 mutants (Fig. 3C).

In both S. cerevisiae and in N. crassa, Prm1p (PRM1) function is required in both fusion pair partners (Heiman and Walter, 2000; Fleissner et al., 2009a). For example, in N. crassa, membrane fusion frequency, which is near 100% in adhered wild type germlings, is reduced to ∼ 75% in pairings between wild type + ΔPrm1, an intermediate value as compared with ΔPrm1 + ΔPrm1 germlings (∼ 50%; Fleissner et al., 2009a). We therefore assessed whether lfd-1 was also required in both fusion partners. In pairings between wild type and Δlfd-1, an intermediate value of cell fusion (87%) was observed in adhered germlings (Fig. 2C). These data indicate that, similar to what was observed for PRM1 in S. cerevisiae and in N. crassa, lfd-1 is required in both germling fusion partners for wild type levels of membrane merger.

Δlfd-1 mutants are unaffected in chemotropic interactions or oscillation of MAK2 to fusion tips

To assess whether chemotropic interactions were also altered in the Δlfd-1 mutant, we determined the number of Δlfd-1 germlings undergoing directed growth in comparison to wild type germlings. In wild type, 86% (± 3 SD) of germlings showed chemotropic interactions 5 h post germination. Among Δlfd-1 germlings, 88% (± 4 SD) showed directed growth. Thus, the Δlfd-1 phenotype appears to be restricted to events following chemotropic interactions and cell adhesion. To further support this finding, we evaluated the localization of the MAP kinase, MAK2 (MAK2–GFP), during chemotropic interactions in the Δlfd-1 mutant in comparison to wild type germlings. MAK2 is essential for chemotropic interactions between N. crassa germlings (Pandey et al., 2004; Fleissner et al., 2009b). In wild type germlings, MAK2 oscillates from germling tip (conidial anastomosis tubes or CATs) to the cytoplasm every ∼ 8 min; communicating germling pairs show opposing oscillation of MAK2 to CAT tips (Fleissner et al., 2009b). As shown in Fig. 3D, MAK2–GFP showed oscillation to CAT tips in Δlfd-1 (mak-2–gfp) germling pairs undergoing chemotropic interactions, with indistinguishable dynamics from wild type (mak-2–gfp) germlings. These data indicate that lfd-1 is dispensable for cell–cell communication and chemotropic interactions during germling fusion.

Δlfd-1 mutants are blocked at membrane merger

Membrane invaginations were prevalent at the fusion site between adhered Δlfd-1 germling pairs (Fig. 2) suggesting that Δlfd-1 mutants are blocked in plasma membrane merger. To test this hypothesis, we compared the region of interaction in wild type germling pairs versus ΔPrm1 or Δlfd-1 germling pairs by transmission electron microscopy (TEM). In wild type germling pairs, apposed plasma membranes in regions of contact were never observed (Fig. 4A). As reported previously (Fleissner et al., 2009a), extensive regions of juxtaposed plasma membranes at contact sites were observed for ΔPrm1 germling fusion pairs (Fig. 4B, inset). The phenotype in the contact zone between Δlfd-1 germlings was remarkably similar to that of ΔPrm1 germling fusion pairs and showed extensive apposed plasma membranes at the contact site, with no intervening cell wall material (Fig. 4C, inset). These data indicate that Δlfd-1 germlings are defective in the step of plasma membrane merger during cell–cell fusion.

Figure 4.

Δlfd-1 shows defects in membrane fusion. Transmission electron microscopy of contact areas between wild type (A), ΔPrm1 (B) or Δlfd-1 (C) germling fusion pairs. Insets show enlargement of contact site between germlings. Scale bars = 0.5 μm.

Δlfd-1 mutants show reduced mating cell fusion during fertilization

In filamentous ascomycete species, mating cell fusion takes place between a specialized hypha (trichogyne) produced by the female reproductive structure and a male cell (Coppin et al., 1997). Fertilization is followed by migration of the male nucleus through the trichogyne and into the female reproductive structure. In N. crassa, karyogamy and meiosis occur ≈ 4 days post fertilization (Raju, 1980). As lfd-1 is required for wild type levels of germling fusion, we hypothesized that lfd-1 may also be involved in the mating cell fusion between a trichogyne and a conidium during fertilization. Such a result would indicate that LFD1 is an integral part of the cell–cell membrane fusion machinery. To test this hypothesis, we performed trichogyne-male cell fusion assays.

Trichogynes sense a spore of the opposite mating type via pheromone/receptor signalling, resulting in chemotropic growth toward the male cell and subsequent cell fusion (Bistis, 1981; Kim and Borkovich, 2006). In our trichogyne-male cell fusion assay, microconidia containing a single histone H1–GFP-labelled nucleus were used as males. Successful fertilization between the trichogyne and the microconidium was detected by a disappearance of the H1–GFP nucleus from the male cell as a consequence of its migration down the trichogyne and into the female reproductive structure (Fleissner et al., 2005; Fig. 5A). We evaluated 50 trichogyne/microconidium interactions from wild type crosses and 50 trichogyne/microconidium interactions in Δlfd-1 homozygous crosses. In the wild type × wild type control crosses, 87% ± 3.25 SD of the trichogyne–microconidium pairs showed loss of H1–GFP fluorescence as a consequence of fertilization (Fig. 5B and D). In contrast, Δlfd-1 X Δlfd-1 homozygous crosses showed a significant reduction of trichogyne–microconidium pairs undergoing successful fusion events (63.5% ± 3.54; P < 0.05; Fig. 5C and D). The reduction in sexual fusion in homozygous Δlfd-1 crosses (≈ 23.5%) was comparable to the reduction in germling fusion observed between Δlfd-1 pairs (≈ 23%). This result supports the hypothesis that LFD1 is part of the cell fusion machinery in N. crassa.

Figure 5.

Δlfd-1 shows fusion defects during mating. Images show female trichogyne attraction and growth toward a male microconidium of the opposite mating type, which bears histone H1–GFP.

A. A cartoon side view of the trichogyne-conidium fusion assay (see Experimental procedures) T = trichogyne; P = protoperithecium (female reproductive structure); M = microcondium. Successful fertilization in wild type × wild type pairings is monitored by the disappearance of the H1–GFP nucleus (green) from the male cell (microconidia, orange) as a consequence of cell fusion and nuclear transport down the trichogyne and into the protoperithecium.

B. Top view of a successful fertilization in a wild type × wild type pairing (FGSC 988 X R11-03).

C. Failure of fertilization in Δlfd-1 X Δlfd-1 (JPG20 X JPG21) trichogyne–conidium interactions, as detected by the persistence of the H1–GFP signal in the male cell. White arrows point at male H1–GFP nuclei that have been targeted by a female trichogyne. Scale bars = 5 μm.

D. Trichogyne-conidium fusion frequencies in wild type × wild type crosses versus Δlfd-1 X Δlfd-1 crosses. Graph bars = SD. Asterisk indicates significant difference with respect to WT (t-test P < 0.05, n = 3).

LFD1 localizes to plasma membrane and endomembranes

lfd-1 is predicted to encode a TM protein. Since mutations in lfd-1 reduced cell fusion frequency, we predicted that LFD1 would localize to the plasma membrane at the point of contact between germling pairs. An lfd-1–gfp allele was constructed by modification of the endogenous lfd-1 locus (JPG22), but fluorescence was very low (data not shown). We therefore constructed a dsRed-lfd-1 allele, expressed under the ccg-1 promoter, and which was integrated at the his-3 locus (Margolin et al., 1997) in a Δlfd-1 his-3 strain (JPG23); full complementation of germling fusion defects was observed (Fig. 2C). dsRED-LFD1 localization during germling fusion was evaluated using live cell imaging and confocal microscopy, which revealed endomembrane localization (Fig. 6A). Similar LFD1 localization was observed using an lfd1–gfp-tagged allele under regulation of a ccg-1 promoter (JPG24) or a tef-1 promoter (JPG29, Fig. S1), both of which also complemented fusion defect in the Δlfd-1 mutant. Similar to fluorescence microscopy results, sucrose gradient cell fractionation of five-hour-old lfd-1–gfp germlings (JPG29) followed by Western blotting using anti-GFP antibodies showed enrichment of LFD1 in fractions corresponding to endomembranes (Fig. 6C). The endomembrane localization of LFD1–GFP in germlings was very similar to that reported for PRM1–GFP (Fleissner et al., 2009a; Fig. 6B). Similar to LFD1, subcellular fractionation of PRM1–GFP from Prm1–gfp 5 h germlings also showed enrichment in the endomembrane fraction (Fig. 6C). Indeed, heterokaryotic germlings carrying both dsRED-LFD1 and PRM1–GFP (JPG23 + JPG28) showed colocalization to endomembranes (Fig. 6B).

Figure 6.

LFD1 localization in germlings.

A. dsRED-LFD1 localization in a fusing Δlfd-1 (ccg-1-dsRed-lfd-1) germling pair.

B. dsRED-LFD1 and PRM1–GFP colocalization in a fusing heterokaryotic germling pair. Scale bar = 5 μm. Arrow shows fusion point.

C. Sucrose gradient fractionation of five-hour-old germlings expressing either LFD1–GFP (top panel) or PRM1–GFP (bottom panel). Fractions were collected and analysed by immunoblot using anti-GFP antibodies (Roche). As controls for fractionation, antibodies to the N. crassa plasma membrane ATPase (PMA1), which is enriched in the plasma membrane fractions, but also localizes to tubular vacuoles (Fajardo-Somera et al., 2013) and antibodies to ERV25, which is involved in ER-Golgi trafficking (T. Starr, unpublished results) were used. Arrows show specific bands (LFD1, PMA1, ERV25 and PRM1), while asterisks show non-specific bands.

Endomembranes in N. crassa consist of endoplasmic reticulum (ER, which shows localization to the nuclear membrane), Golgi, endosomes and vacuoles. To assess endomembrane localization of LFD1 and PRM1, we constructed heterokaryons between LFD1–GFP or PRM1–GFP (JPG29 and JPG30), with strains expressing dsRED(RFP)-tagged proteins developed for subcellular compartments in N. crassa, which included markers for Golgi (RFP-VPS52), endosomes (RFP-RAB4) and vacuoles (RFP-VAM3; Bowman et al., 2009). For both LFD1–GFP and PRM1–GFP, we observed colocalization with the vacuolar marker RFP-VAM3 (Fig. S2).

Although localization of PRM1–GFP to germling contact points was not previously observed, localization to the plasma membrane in hyphae was reported (Fleissner et al., 2009a). We therefore evaluated localization of dsRED-LFD1 within the interconnected hyphal network that makes up a N. crassa colony. Localization of dsRED-LFD1 to endomembranes was observed in hyphae throughout the colony (Fig. 7). LFD1 localization to the plasma membrane was absent in apical, leading hyphae at the edges of the colony (which do not undergo fusion, Hickey et al., 2002; Bistis et al., 2003; Fig. 7A and D) and from most trunk hyphae in the interior of the colony (Fig. 7B and D). However, plasma membrane localization was frequently observed in fusion hyphae in the interior of the colony, and which are defined by having a reduced hyphal diameter, as compared with trunk and apical hyphae (Steele and Trinci, 1975; Bistis et al., 2003; Simonin et al., 2012; Fig. 7C and D). LFD1 localization in this region was particularly evident in the region of the plasma membrane associated with septa (Fig. 7C and D).

Figure 7.

Localization of LFD1 in a fungal colony.

A. dsRED-LFD1 localization in apical hyphae at the periphery of a colony.

B. dsRED-LFD1 localization in trunk hyphae in the interior of a colony.

C. dsRED-LFD1 localization in fusion hyphae in the interior of a colony. Scale bar = 5 μm. Arrows shows septal and plasma membrane localization of LFD1.

D. Apical and interior trunk hyphae can be distinguished from fusion hyphae by hyphal diameter (Steele and Trinci, 1975; Bistis et al., 2003; Simonin et al., 2012) (left panel). Frequency of plasma membrane localization of LFD1 in different hyphae (right panel). Graph bars = SD.

The similarity in phenotype between ΔPrm1 and Δlfd-1 mutants in N. crassa suggested that these proteins function in a similar process during cell fusion. To test this hypothesis, we performed an epistasis test by constructing Δlfd-1; ΔPrm1 double mutants. Compared with the Δlfd-1 and ΔPrm1 fusion frequencies in adhered germlings (77% and 52% respectively), the Δlfd-1; ΔPrm1 germlings (JPG26 + JPG27) showed a significant reduction in fusion frequency (29%; t-test P < 0.05; Fig. 8). Furthermore, the Δlfd-1; ΔPrm1 double mutant showed a reduction in aerial hyphae extension (Fig. 3C), a phenotype typically displayed by hyphal fusion mutants of N. crassa.

Figure 8.

Δlfd-1; ΔPrm1 germlings show a reduction in germling fusion frequency relative to fusion frequencies in wild type, Δlfd-1 or ΔPrm1 germlings. Δlfd-1; ΔPrm1 germling fusion frequency versus individual mutants and wild type. Bars = SD, n = 3.

Cell lysis during germling fusion in Δlfd-1 and ΔPrm1 mutants is affected by extracellular calcium concentration

In S. cerevisiae, a reduction of extracellular Ca2+ concentration induces an increase in cell lysis in prm1Δ and fig1Δ mating cells (Aguilar et al., 2007). In N. crassa, chemotropic interactions and cell fusion are abolished in the absence of extracellular Ca2+ (Palma-Guerrero et al., 2013). We therefore evaluated chemotropic interactions and fusion frequencies in wild type and ΔPrm1 germlings in media containing varying levels of Ca2+. At an extracellular concentration of 0.34 mM Ca2+ (reduced Ca2+) chemotropic interactions and cell fusion in wild type germlings was normal (Fig. 9). In ΔPrm1 germlings, a decrease in cell fusion, which was associated with cell lysis, was observed at this Ca2+ concentration, although chemotropic interactions were normal.

Figure 9.

Extracellular calcium concentration affects membrane fusion and cell lysis in germling pairs.

A. Micrograph of germling fusion in a wild type pair (left) versus lysed fusion pairs (right). Cells were stained with 0.002 % methylene blue. Scale bars = 5 μm.

B. Frequency of germling adherence (non-fused), fused germlings (fused) and pairs that underwent lysis (lysed) in wild type, ΔPrm1, Δlfd-1 and Δlfd-1; ΔPrm1 germlings under calcium replete (normal) versus reduced Ca2+ concentration (reduced).

C. Frequency of germling adherence, fusion and lysis in Δlfd-1, ΔPrm1 and Δlfd-1; ΔPrm1 germlings at different Ca2+ concentrations.

D. Frequency of germling adherence, fusion and lysis in ΔPrm1, Δlfd-1 and Δlfd-1; ΔPrm1 germlings and when mixed with wild type germlings at normal and reduced Ca2+ concentration. Graph bars = SD, n = 3.

To differentiate the frequency of cell lysis versus cell fusion in wild type versus the mutants, we assessed the phenotype of germling pairs showing cell–cell adherence by treating with the vital dye, methylene blue (Suzuki et al., 2000; Fig. 9A). In both ΔPrm1 and Δlfd-1 germling pairs, we observed three phenotypic classes under normal Ca2+ concentrations: adhered, but non-fused; adhered, but showing lysis of both germlings in a pair; adhered and fused (Fig. 9A and B). In both the ΔPrm1 and Δlfd-1 strains, a significant proportion of germlings showed cell lysis (Fig. 9B). In addition to methylene blue staining, we often observed stained cytoplasmic contents leaking out of ΔPrm1 and Δlfd-1 germling pairs, indicating plasma membrane damage.

The cell lysis phenotype was exacerbated by a reduction in extracellular Ca2+ in both the ΔPrm1 and Δlfd-1 germling pairs, but was not affected by osmotic stabilization (data now shown). In particular, the Δlfd-1; ΔPrm1 double mutant showed a large increase in cell lysis under reduced Ca2+ conditions (∼ 70% of germling pairs lysed). Increasing extracellular Ca2+ concentration suppressed the lysis phenotype in all three strains, increasing fusion frequencies among ΔPrm1, Δlfd-1 and Δlfd-1; ΔPrm1 adhered germling pairs (Fig. 9C). Unlike the bilateral cell fusion phenotype, the presence of PRM1 or LFD1 in the membrane of one of the cells of a fusion pair was sufficient to suppress cell lysis under reduced Ca2+ conditions in ΔPrm1 and Δlfd-1 germlings (Fig. 9D).

The cell lysis phenotype of adhered ΔPrm1 or Δlfd-1 germling fusion pairs prompted us to test the lysis phenotype of strains carrying deletions for the other five candidate genes (NCU05798, NCU05629, NCU05681, NCU03863 and NCU06912), which were selected in the initial screening for germling fusion defects after treatment with FM4–64. Interestingly, germlings of all five of these deletion strains showed an increase in cell lysis in adhered cells in comparison to wild type germlings (Fig. S3). However, the cell lysis phenotype was not sensitive to external Ca2+ concentrations; deletion mutants showed no difference in cell lysis frequency between normal and reduced Ca2+ concentrations. In particular, germlings undergoing fusion from a strain carrying a deletion of NCU06912 (encoding a hypothetical protein predicted to be secreted) showed a ∼ 20% cell lysis phenotype under both normal and reduced Ca2+ conditions, a value similar to both ΔPrm1 and Δlfd-1 mutants under reduced Ca2+ conditions.

LFD1 and PRM1 have redundant function

Our genetic data indicated that lfd-1 and Prm1 have additive functions during membrane merger and cell fusion. We therefore hypothesized that overexpression of one member may compensate for the loss of the other member, resulting in a decrease in lysis and an increase in cell fusion frequencies. To test this hypothesis, we overexpressed Prm1 in a Δlfd-1 mutant and overexpressed lfd-1 in a ΔPrm1 mutant, using the promoter of the tef-1 gene, which is highly expressed in germlings (Dettmann et al., 2012); high expression levels of the Prm1 and lfd-1 constructs were confirmed by RT-PCR (Fig. S4). Both the tef-1-Prm1 and tef-1-lfd-1 constructs restored wild type germling fusion frequency in ΔPrm1 and Δlfd-1 mutants respectively (JPG30 and JPG29). Consistent with our hypothesis, overexpression of Prm1 in Δlfd-1 (JPG31) cells suppressed the Δlfd-1 phenotype, such that fusion and cell lysis frequencies were similar to Δlfd-1 + wild type pairings (Fig. 10A). Similarly, overexpression of lfd-1 in a ΔPrm1 (JPG32) mutant restored fusion and cell lysis frequencies, resembling wild type + ΔPrm1 pairings (Fig. 10B).

Figure 10.

LFD1 and PRM1 have redundant functions.

A. Frequency of germling adherence, fusion and lysis in a Δlfd-1 mutant when paired with Δlfd-1 (tef-1-lfd-1) or Δlfd-1 (tef-1-Prm1) germlings.

B. Adherence, fusion and lysis phenotype of ΔPrm1 germlings paired with ΔPrm1 (tef-1-Prm1) or ΔPrm1 (tef-1-lfd-1) germlings. Underlined genes indicated overexpression via regulation by the tef-1 promoter. Graph bars = SD, n = 3.

Lysis is a result of a membrane disruption during defective membrane fusion

The increased cell lysis phenotype observed in the ΔPrm1 and Δlfd-1 germlings under reduced Ca2+ concentration suggests that lysis is a result of membrane damage produced during the cell fusion process. This hypothesis fits with a model for membrane damage repair, which requires extracellular Ca2+ (Yawo and Kuno, 1985; Steinhardt et al., 1994). In mammalian cells, repair of membrane damage has been proposed to be mediated by Ca2+-triggered exocytosis of lysosomes, with synaptotagmin VII as a potential Ca2+ sensor (Reddy et al., 2001). In S. cerevisiae, a mutant bearing deletions of PRM1 and a synaptotagmin homologue, TCB3, shows cell lysis levels during mating that are similar to those observed for prm1Δ cells in the absence of Ca2+. These observations suggest that Tcb3p plays a role in membrane repair during defective membrane fusion in mating cells (Aguilar et al., 2007). To test the hypothesis that Ca2+-dependent membrane repair during germling fusion in N. crassa requires proteins similar to synaptotagmin, we screened the N. crassa genome for genes homologous to TCB3. The encoded product of NCU03263 showed homology to Tcb3p (30% identity, E-value = 0) and to mammalian synaptotagmin I (24% identity, E-value = 2e–25), which is a Ca2+ sensor for neurotransmitter release in neuronal cells (Xu et al., 2007). We named NCU03263 gene syt-1 for synaptotagmin-like-1.

SYT1 is predicted to contain 1 TM domain and 4 C2 domains, which are involved in targeting proteins to the cell membrane (Cho, 2001). The Δsyt-1 mutant itself (JPG33 + JPG34) showed a wild type growth and normal chemotropic interactions and germling fusion phenotype, although the slight, but not significant (t-test, P > 0.05), increase in cell lysis in wild type adhered germlings under reduced Ca2+ conditions was recapitulated in Δsyt-1 mutants in Ca2+ replete conditions (Fig. 11A and B). The Δsyt-1 Δlfd-1 (JPG35 + JPG36) and Δsyt-1; ΔPrm1 (JPG37 + JPG38) mutants showed a significant (t-test, P < 0.05) increase in cell lysis under Ca2+ replete conditions that was very similar to Δlfd-1 or ΔPrm1 mutants under conditions of reduced Ca2+ (Fig. 11A and B). Furthermore, a decrease in cell lysis was not observed in the Δsyt-1 Δlfd-1 and Δsyt-1; ΔPrm1 mutants by increasing extracellular Ca2+ concentrations. These results support a role for SYT1 and calcium in the prevention of cell lysis during defective membrane fusion in Δlfd-1 and ΔPrm1 germlings. Furthermore, the fact that both Δsyt-1 Δlfd-1 and Δsyt-1; ΔPrm1 mutants show the similar values of fusion and lysis under different calcium concentrations suggests that no other proteins are involved in the calcium-dependent membrane repair mechanisms during cell fusion in N. crassa, apart from SYT1.

Figure 11.

Localization and role of SYT1 during germling fusion. Both Δlfd-1 and ΔPrm1 mutants need SYT1 and extracellular Ca2+ to prevent cell lysis during defective membrane fusion.

A. Δsyt-1 Δlfd-1 germling adherence, lysis and fusion at different Ca2+ concentrations as compared with wild type, Δsyt-1 or Δlfd-1 germlings.

B. The adherence, fusion or lysis phenotype of Δsyt-1; ΔPrm1 germlings at different Ca2+ concentrations in comparison to wild type, Δsyt-1 or ΔPrm1 germlings. Graph bars = SD, n = 3.

C. SYT1–GFP localization in ungerminated conidia (a), in germlings (b), in germlings undergoing chemotropic interactions (c) and in fusing germlings (d). Arrow indicates point of fusion. Scale bar = 5 μm.

In S. cerevisiae, Tcb3p has recently been shown to be involved in ER-PM tethering and localizes to the cortical ER (Manford et al., 2012). We assessed localization of SYT1 in N. crassa by constructing a syt-1–gfp allele and assessing SYT1–GFP localization in wild type (JPG39) and Δlfd-1 (JPG40) mutants. Full functionality of the syt-1–gfp construct was confirmed by complementation of the lysis phenotype of a Δsyt-1; ΔPrm1 (syt-1–gfp) strain (JPG41). The adhered germling lysis frequency of the Δsyt-1; ΔPrm1 (syt-1–gfp) strain resembled the values of the ΔPrm1 mutant (14.66% ± 4.16 SD versus 11.11% ± 1.92 SD respectively), significantly lower than the frequency observed for a Δsyt-1; ΔPrm1 strain (29.33% ± 2.3SD). In wild type cells, SYT1–GFP showed a primarily plasma membrane-associated localization in puncta around the periphery of ungerminated conidia (Fig. 11C, panel a), a localization pattern very similar to that of both Tcb3p and Prm1p (in unstimulated cells) in S. cerevisiae (Heiman and Walter, 2000; Manford et al., 2012). In germinating conidia and in germlings undergoing chemotropic interactions, SYT1–GFP localized as a plasma membrane-associated crescent at germling tips (Fig. 11, panels b, c). Upon cell fusion, SYT1–GFP showed more prominent endomembrane localization, as well as plasma-membrane associated puncta.

To assess dynamic localization of SYT1–GFP during cell fusion in wild type and Δlfd-1 germlings, we performed live cell imaging of SYT1–GFP localization during germling fusion. SYT1–GFP localized to plasma-membrane associated puncta, endomembranes, and as a crescent in germling tips during chemotropic interactions. Upon cell fusion, SYT1–GFP disappears from the contact site, but plasma-membrane associated SYT1–GFP remained on either side of the fusion pore. No obvious differences were detected in SYT1–GFP localization during chemotropic interactions and cell fusion between wild type germlings (Video S1) or during chemotropic interactions and cell fusion between Δlfd-1 germlings (Video S2).

Discussion

Previously, PRM1 was the only protein known to act at the plasma membrane fusion step in N. crassa (Fleissner et al., 2009a). Here, we provide evidence of a new player with the identification of LFD1. The Δlfd-1 mutant is similar in phenotype to a ΔPrm1 mutant, with defects in both germling and trichogyne-conidium fusion (Fleissner et al., 2009a). Ultrastructural analyses of ΔPrm1 and Δlfd-1 germlings undergoing somatic cell fusion showed juxtaposed plasma membranes and invaginations, a phenotype never observed in contact areas between wild type germling pairs. These data support a role for both PRM1 and LFD1 in the machinery involved in plasma membrane merger in N. crassa.

Cell lysis and the role of calcium during membrane fusion

Apart from the membrane fusion defect, another phenotype shared between Δlfd-1 and ΔPrm1 is calcium-responsive cell lysis, a phenotype also observed in S. cerevisiae prm1Δ mating pairs (Aguilar et al., 2007). A second TM protein, Fig1p, is also implicated in membrane fusion during mating in S. cerevisiae. In both N. crassa and S. cerevisiae, Fig1 has a role in the low-affinity calcium uptake system (Muller et al., 2003; Cavinder and Trail, 2012). The N. crassa Δfig-1 mutant shows a mating-type specific reproductive defect (Cavinder and Trail, 2012), but its role in germling fusion is unknown. Our preliminary data suggests that the primary defect of the Δfig-1 mutation in adhered germlings is a slight increase in cell lysis (data not shown). Further work will reveal the relationship between PRM1, FIG1 and LFD1 in germling fusion, membrane merger and cell lysis in N. crassa.

Our genetic data indicate that LFD1 and PRM1 function in compensatory pathways during membrane merger in N. crassa. The loss of both Δlfd-1 and ΔPrm1 exacerbated the membrane fusion and cell lysis phenotype, with up to 70% of adhered germlings undergoing lysis under conditions of reduced calcium. The increase in cell lysis was correlated with a decrease in fused germling pairs (Fig. 9), but also an increase in lysis in germling pairs that showed adherence. This phenotype of the Δlfd-1; ΔPrm1 mutant was in contrast to the phenotype of the Δlfd-1 or ΔPrm1 germlings on reduced calcium, where an increase in cell lysis was correlated with a decrease in cell fusion, but the percentage of adhered, non-fused germlings remained the same as in calcium replete conditions. These data suggest that in the absence of both Δlfd-1 and ΔPrm1, and under reduced calcium, adhered germlings that have engaged the fusion machinery, but that have not initiated membrane pore formation, become susceptible to membrane damage/cell lysis. Importantly, the overexpression of lfd-1 in ΔPrm1 mutant and overexpression of Prm1 in a Δlfd-1 mutant complemented the fusion and cell lysis phenotypes, suggesting these two proteins mediate a similar process during membrane merger.

It was previously suggested for S. cerevisiae that cell lysis during mating is a direct result of engagement of the cell fusion machinery and is intrinsically linked to the mechanism of lipid bilayer fusion (Jin et al., 2004; Aguilar et al., 2007). Membrane rupture and cell lysis during the fusion process may be ameliorated by accessory proteins, such as Prm1p and Fig1p in S. cerevisiae (Aguilar et al., 2007) and PRM1, LFD1 and possibly FIG1 in N. crassa. Alternatively, PRM1 and LFD1 may play a role in organizing the plasma membrane fusion machinery and which is important for membrane integrity, thus explaining the reduced fusion activity in cells lacking PRM1 and LFD1.

In S. cerevisiae, Prm1p–GFP shows ER and endosomal localization (Heiman and Walter, 2000; Olmo and Grote, 2010), although after α-factor exposure, a fraction of Prm1p localizes to the shmoo tip and points of contact between mating cells. In N. crassa, PRM1–GFP localized to endomembranes in germlings and to endomembranes and the plasma membrane in hyphae (Fleissner et al., 2009a). An identical localization pattern was detected here for LFD1. In particular, plasma membrane localization of LFD1 in a colony was enriched in hyphae that often engage in fusion events (Fig. 7). In S. cerevisiae, extremely low concentrations of Prm1p are sufficient to promote membrane merger (Olmo and Grote, 2010). These data suggest that, similar to S. cerevisiae Prm1p, LFD1 may be transiently recruited to the point of cell contact between germlings during membrane merger. This hypothesis is supported by the presence of a dileucine motif in LFD1, an endosomal-sorting signal that can act as endocytosis internalization signal for membrane proteins and receptors, and is involved in the recycling of plasma membrane proteins (Pandey, 2010).

Synaptotagmin plays an important role in membrane repair during cell–cell fusion

An influx of Ca2+ occurs during mating in S. cerevisiae and is facilitated by activation of either the high affinity calcium transport system or the low affinity calcium transport system (Iida et al., 1994; Muller et al., 2003). We predict that Ca2+ influx is also an important process in N. crassa, both for chemotropic interactions (Palma-Guerrero et al., 2013) and for membrane merger/repair. Here we provide evidence that syt-1, a N. crassa synaptotagmin homologue, may be involved in membrane damage repair during germling fusion. Strains carrying deletions of syt-1syt-1 Δlfd-1 and Δsyt-1; ΔPrm1) phenocopy the increased lysis observed upon Ca2+ depletion in ΔPrm1 and Δlfd-1 adhered germlings. S. cerevisiae has three synaptotagmin-like proteins: Tcb1p, Tcb2p and Tcb3p (tricalbins; Creutz et al., 2004). In N. crassa, only one homologue to TCB1/TCB2/TCB3 is present: NCU03263 (syt-1). In S. cerevisiae, the tricalbins localize to the cortical ER and have been implicated in ER-PM tethering (Manford et al., 2012). In N. crassa, SYT1 localized to a plasma-membrane associated crescent in germlings and as plasma membrane associated puncta in asexual spores, presumably marking cortical ER in these cells. Cortical or ‘peripheral’ ER has been described in Ustilago maydis and Aspergillus nidulans, primarily based on localization of known ER components (Wedlich-Soldner et al., 2002; Markina-Inarrairaegui et al., 2013). The Δsyt-1; ΔPrm1 and Δsyt-1 Δlfd-1 mutants showed similar levels of fusion and lysis under reduced or increased calcium, suggesting that SYT1 is the only synaptotagmin-like protein involved in membrane repair during germling fusion in N. crassa. Our characterization of syt-1 is the first evidence of a possible membrane repair mechanism identified in a filamentous fungus.

The unknown role of genes in which deletion results in a strain showing calcium-independent cell lysis during fusion

Apart from lfd-1, we identified five genes regulated by PP1 in which deletion strains show a cell lysis phenotype in adhered germlings; cell lysis in these five deletion strains was not observed in the absence of cell contact. The frequency of lysis of adhered germling pairs in these five deletion strains was unaffected by extracellular Ca2+ concentration. Four genes encode hypothetical proteins and homologues were only identified in genomes of other filamentous ascomycete species. One gene, NCU03863, encodes a protein with homology to Rbs1p, a sphingoid transporter in S. cerevisiae. Rsb1p is proposed to be involved in establishment of plasma membrane lipid asymmetry by catalysing an ATP-dependent translocation of sphingoid long-chain bases from the cytoplasmic to extracytoplasmic face of the membrane (Kihara and Igarashi, 2002); a cell lysis phenotype for rsb1Δ has not been reported. These observations suggest that an abnormal distribution of lipids in the plasma membrane contact area in ΔNCU03863 germling pairs may result in membrane damage and lysis during the fusion process. The lipid composition of a membrane has a large influence on the biophysical properties that may affect the ability of membranes to fuse (Kasson and Pande, 2007). The fungal plasma membrane is highly enriched in ergosterol, similar to the animal sterol, cholesterol, but with some structural differences (Parks and Casey, 1995). Within a membrane, sterols and sphingolipids cluster into liquid-ordered domains, called lipid rafts. In addition to lipid rafts, the inner and outer leaflets of the plasma membrane are also distinct in lipid composition and lipid asymmetry is maintained by active transport of lipids (Alvarez et al., 2007). Furthermore, in S. cerevisiae, the sterol content of the plasma membrane affects its propensity to undergo fusion via a Prm1-regulated protein complex. In the absence of Prm1, membrane fusion was completely absent in ergosterol-depleted membranes (Jin et al., 2008). It has also been reported that the structure of sterol aliphatic chains affects yeast cell shape and cell fusion during mating (Aguilar et al., 2010). Although sphingolipids have a more important role in pheromone signalling in S. cerevisiae than in plasma membrane fusion, depletion of both ergosterol and sphingolipids was a requirement for a low level of ergosterol–sphingolipid interaction needed for membrane fusion (Jin et al., 2008). These observations suggest a sophisticated complexity regarding structural requirements of sterols and sphingolipids for cell fusion. Further investigation into the molecular function of proteins encoded by the five genes showing cell lysis during cell fusion in N. crassa, and their relationship to PRM1 and LFD1 during plasma membrane merger, promises to reveal important and novel aspects associated with membrane fusion pore formation and repair.

Experimental procedures

Strain and plasmid construction

The deletion mutants used in this study were generated by the Neurospora Genome Project (Dunlap et al., 2007), and obtained from the Fungal Genetics Stock Center (FGSC; McCluskey, 2003). Strains used in this study are listed in Table S1.

To obtain double mutants with ΔPrm1, crosses were performed with a ΔPrm1 strain complemented at the his-3 locus with the ccg1–Prm1–gfp (A24) since otherwise this strain is sterile due to dominant post-fertilization defects in the ΔPrm1 mutant. Segregation analyses showed a secondary mutation occurred in the ΔNCU09307 strain (FGSC 20299) that affected the chemotropic phenotype. This secondary mutation was removed via segregation analyses; the hygromycin marker (ΔNCU09307::hph) co-segregated with the membrane invaginations/cell fusion defect.

The his-3 A strain (FGSC 6103) and a his-3 a strain (FGSC 9716) were used as males in crosses with deletion mutants. Progeny carrying the deletion mutation and the his-3 marker were used for transformation, using either pMF272 or pMF334 (Freitag et al., 2004; Freitag and Selker, 2005) by electroporation of macroconidia as described by (Margolin et al., 1997). The plasmid for cytoplasmic mCherry was obtained by replacing sGFP in pMF272 by mCherry from plasmid pMFP26 (kindly provided by Dr S. Brody, University of California, San Diego). To obtain histone H1–GFP transformants, his-3 strains were transformed with pMF280 (Freitag et al., 2004).

Modification of the endogenous lfd-1 locus for lfd-1–gfp constructs was achieved by fusion PCR-amplification. Primers (Table S2) were designed to amplify a 1.3 Kb fragment at the 3′ flank of the gene, excluding the stop codon (9307Fup and 9307Rup), and a 1.3 Kb fragment after the stop codon (9307Fdown and 9307Rdown), as well as a fragment containing GFP and the hygromycin cassette from the vector pGFP::hph::loxP (GFP-F and hyg-R; Honda and Selker, 2009). The three fragments were fused by nested PCR using primers 9307Fnested and 9307Rnested (Table S2). The PCR fragment was TOPO (Invitrogen) cloned, amplified by PCR, and transformed in Δmus-52::barR strain for homologous recombination with the endogenous lfd-1 locus, as described (Colot et al., 2006). Positive transformants growing on hygromycin B (200 μg ml−1; Staben et al., 1989) were backcrossed with wild type to remove the Δmus-52::barR mutation. The correct genotype was confirmed by PCR and the mus-52 background was confirmed by ignite selection (Hays and Selker, 2000).

For subcellular localization of LFD1 and SYT1, plasmids containing either gfp or dsRed fusion constructs under the ccg-1 promoter were built based on the plasmids pMF272 and PMF334. The ORFs of lfd-1 and syt-1 were amplified by PCR as annotated by the N. crassa database (http://www.broadinstitute.org/annotation/genome/neurospora/MultiHome.html) using genomic DNA with primers that incorporated XbaI/PacI and SpeI/XbaI sites, respectively, at the ends (Table S2), and introduced into pMF272 or pMF334. The plasmid containing the lfd-1 gene under the tef-1 promoter was constructed by replacing Prm1 by lfd-1 into the Tef-1–Prm1–GFP plasmid kindly provided by Dr A. Fleissner (Technische Universitat Braunschweig, Germany) using the XbaI/PacI sites.

To obtain ectopically expressed SYT1–GFP, the fragment containing ccg1–GFP, and the multiple cloning site in between ccg1 and GFP, was digested from pMF272 and cloned into pBC-Phleo (Silar, 1995) using NotI/EcoRI sites. The open reading frame for syt-1 was introduced into this vector by using XbaI/PacI at the multiple cloning site, generating ccg1-syt-1–GFP in pBC-Phleo. Positive transformants were selected on phleomycin (10 ng ml−1).

Growth conditions and cell lysis assays

All strains were grown on Vogel's medium (Vogel, 1956) and all crosses were performed on Westergaard's synthetic cross medium (Westergaard and Mitchell, 1947). Growth rates were measured in race tubes by measuring linear growth in mm from 24 to 72 h at 25°C in constant light, in triplicate for each strain. Aerial hyphae extension was measured from tubes containing 3 ml of liquid VMM inoculated with 106 conidia after 3 days of growth at 25°C in constant light. Ten replicates were made for each strain.

For lysis experiments in reduced Ca2+, Vogel's MM containing 0.34 mM final concentration of CaCl2·2H2O was used. Germlings were prepared and examined as above, but 0.002% methylene blue was applied to the agar blocks before imaging to differentiate lysed from non-lysed cells. The same procedure was used to evaluate lysis under increased calcium concentrations (2× = 1.28 mM and 5× = 3.2 mM).

Membrane fusion screening and fluorescence microscopy

Strains were grown on Vogel's minimal media (VMM; Vogel, 1956) slant tubes for 4–6 days or until significant conidiation occurred. Conidial suspensions were prepared by collecting conidia with wood sticks and suspended in 600 μl of sterile distilled water and filtered. Conidia were diluted to a concentration of 3 × 107 conidia ml−1 and 300 μl were spread onto a VMM plate. Plates were incubated for 5 h at 30°C and 1 cm squares were excised and treated with 4 μM FM4–64. Samples were observed with a Zeiss Axioskop 2 using a 403 Plan-Neofluor oil immersion objective and a Rhodamine filter set (excitation at 543 nm and emission at 590 nm). Images were captured using a Hamamatsu Digital Camera C4742-95 (Hamamatsu, Japan) and OpenLab 4.0.3 software.

Strains expressing either cytoplasmic GFP or dsRED (or mCherry) were used to prepare conidial suspensions as above. Fusion events were counted for germling pairs that expressed fluorescence in both red and green channels. Images were taken using a Leica SD6000 confocal microscope with a 100 × 1.4 NA oil-immersion objective equipped with a Yokogawa CSU-X1 spinning disk head and 488-nm and 561 lasers controlled by Metamorph® software (Molecular Devices).

Protein localization was analysed in conidial germlings prepared as described above, or in hyphal samples growing in VMM agar plates. Heterokaryotic cells for the different tagged proteins were obtained by mixing equal amounts of the two strains expressing the different proteins on VMM slant tubes. After conidiation, conidial suspensions were prepared and examined by confocal microscopy (as above).

Electron microscopy

Transmission electron microscopy was performed as described by Fleissner et al. (2009a), with minor modifications. Three hundred microlitres of conidia at a concentration of 3 × 107 cells ml−1 was spread on Vogel's minimal media plates and incubated at 30°C for 5 h. After the fixation and KMnO4 treatment, as described by Fleissner et al. (2009a), samples were dehydrated through a graded acetone series by incubating for 10 min at each acetone concentration from 35 to 100%. Samples were subsequently infiltrated with epon/araldite resin. Resin was added at the following concentrations for 1 h each: 2:1, 1:1 and 1:2 (acetone : resin). Pure resin was then added and incubated overnight. The following day, fresh pure resin was added along with accelerant and incubated for 4 h. Samples were embedded in moulds for 72 h. Sections of 60–70 nm were cut on a Leica microtome and stained with 2% methanol uranyl acetate and lead citrate. Samples were imaged under an FEI Tecnai 12 transmission electron microscope at 100 kV (Electron Microscopy Lab, UC Berkeley).

Trichogyne-microconidium fusion assays

Trichogyne-conidium fusion assays were performed as described previously (Fleissner et al., 2005). A small piece of mycelium was used to inoculate the centres of plates containing 2% water agar (Bacto agar; Difco). Plates were incubated for 2 days in a humidity chamber and then transferred to the bench top for five more days until protoperithecia (female reproductive structures) had developed. Thin slices (approximately 5 × 5 × 1 mm) of 2% water agar were placed on top of groups of one to five protoperithecia. A water droplet containing five to 100 H1–GFP microconidia was applied as a spot on top of the agar slice. After 24 and 48 h of incubation at room temperature, the plates were evaluated by bright-field and fluorescence microscopy.

RNA extraction and qRT-PCR

RNA was isolated from conidia at a concentration of 1 × 106 cells/ml grown in constant light for 5 h in liquid Vogel's media at 25°C, 2.5 h with shaking at 200 rpm and 2.5 h without shaking. Total RNA from frozen samples was isolated using Zirconia/Silica beads (0.5 mm diameter; Biospec) and a Mini-Beadbeater-96 (Biospec) with 1 ml TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Total RNA was further purified by digestion with TURBO DNA-free (Ambion) and RNA concentration was measured by Nanodrop.

Quantitative RT-PCR was performed using the iScript One-Step RT-PCR Kit with SYBR Green (Bio-Rad) and the CFX Connect Real-Time PCR Detection System (Bio-Rad). Reactions were performed in triplicate with a total reaction volume of 20 μl including 150 nM each forward and reverse primers (Table S3) and 25 ng of template RNA. Data was normalized to the endogenous control actin.

Subcellular fractionation and Western blotting

Conidia were harvested as described and inoculated in 100 ml VMM flasks at a concentration of 106 conidia ml−1. Flasks were incubated 2.5 h at 30°C and shaking at 200 rpm, then 2.5 h at 30°C without shaking. Samples were harvested by vacuum filtration over a PVDF membrane (0.45 μm pore size) and frozen in liquid nitrogen. Sucrose gradients were prepared as described in Kaiser et al. (2002), with modifications. Cells were suspended in 0.5 ml STE10 (10% [wt/wt] sucrose, 10 mM Tris-HCl pH 7.5, 10 mM EDTA) with complete protease inhibitors using 1.5 ml microcentrifuge tubes. Glass beads (0.5 mm diameter, BioSpec Products, Inc.) were added to the meniscus and cells were disrupted by vortexing at top speed for 2 min at 4°C and cleared of unlysed cells and cell walls by centrifugation at 500 g for 5 min at 4°C. 0.3 ml of lysate were carefully layered on top of a 4.5 ml, 20–50% [w/w] sucrose gradient in 10 mM Tris-HCl, pH 7.5, 10 mM EDTA. Samples were centrifuged for 18 h at 4°C at 100 000 g in a SW60 rotor (Beckman Instruments, Fullerton, CA). Eleven 375 μl fractions were collected from the top of the gradient and stored at 20°C, and discarding the last fraction at the bottom of the tube containing the pellet. Fifty microlitres of each fraction was mixed with 50 μl 2× sample buffer + EDTA (4% SDS, 0.1 M Tris-HCl, pH 6.8, 4 mM EDTA, 20% glycerol, 2% 2-mercaptoethanol, 0.2% Bromophenol blue), incubated at 37°C for 1 h to solubilize proteins, and then 45 μl was loaded onto a gel for SDS-PAGE. Samples were run on a 4–12% Nu-Page Bis-Tris Gel (NOVEX, Life Technologies). Protein gels were subjected to Western blot analysis using anti-GFP antibody (1:1000 dilution; Roche). Anti-PMA1 antibody was used as a control for the fractions containing plasma membrane (1:3000 dilution; kindly provided by Dr K. Allen, Yale School of Medicine), and anti-ERV25 antibody was used as a control for the fractions containing ER (1:1000 dilution; kindly provided by Dr T. Starr, Energy Biosciences Institute, University of California Berkeley). Each experiment was repeated twice with identical results.

Acknowledgements

We acknowledge Reena Zalpuri and Kent McDonald of the University of California Berkeley Electron Microscope Laboratory for their assistance in the TEM imaging. We thank Andre Fleissner (Technische Universitat Braunschweig, Germany) for the tef-1 promoter vector, Stuart Brody (University of California, San Diego) for the pMFP26 plasmid, Barry Bowman (University of California, Santa Cruz) for the RFP-VPS-52, RFP-VAM-3 and RFP-RAB4 strains, Trevor Starr (Energy Bioscience Institute, University of California, Berkeley) for replacing sGFP by mCherry in the pMF272 plasmid and for the anti-ERV25 antibody used as a control for the sucrose gradient fractionation, and Kenneth Allen (Yale School of Medicine) for the PMA1 antibody. This work was supported by National Science Foundation Grant MCB 1121311 to N.L.G. We are pleased to acknowledge use of materials generated by NIH Program Project Grant P01GM068087, functional analysis of a model filamentous fungus.

Ancillary