The Predicted Mannosyltransferase GT69-2 Antagonizes RFW-1 To Regulate Cell Fusion in Neurospora crassa

Cell wall remodeling is a dynamic process that balances cell wall integrity versus cell wall dissolution. In filamentous fungi, cell wall dissolution is required for somatic cell fusion and conidial separation during asexual sporulation.

between some strains carrying alternative cwr alleles and cells complete the fusion process (29). However, in screening germinated conidia (germlings) from a Dcwr-1 Dcwr-2 mutant (Dcwr-1 DNCU01381 Dcwr-2) (Table S1 in the supplemental material) against other wild-type N. crassa isolates, we observed that the Dcwr-1 Dcwr-2 mutant failed to undergo cell fusion when paired with wild-type strain JW224 (Fig. 1A), suggesting the existence of a second locus that regulated cell wall dissolution during somatic cell fusion. To identify this second locus, we performed bulk segregant analysis (BSA) of progeny from a cross between FGSC2489 (the parental laboratory strain of the Dcwr-1 Dcwr-2 mutant) and JW224. Progeny segregated into three classes: (i) progeny that underwent chemotropic interactions with FGSC2489 and JW224, but only completed cell fusion with FGSC2489; (ii) progeny that underwent chemotropic interactions with FGSC2489 and JW224, but only completed cell fusion with JW224; and (iii) progeny that failed to fuse with either parent. This third class of progeny was paired with the Dcwr-1 Dcwr-2 mutant; approximately half of these progeny fused with the Dcwr-1 Dcwr-2 strain, while the other approximately half did not. Genomic DNA from these two progeny pools of the third class, one pool of progeny that fused with the Dcwr-1 Dcwr-2 mutant and the second progeny pool that failed to fuse with the Dcwr-1 Dcwr-2 mutant, was isolated and subjected to whole-genome resequencing. From a comparison of single nucleotide polymorphisms (SNPs) between these two pools, a region spanning approximately 3 Mb on chromosome VI was identified that showed SNP segregation between the Dcwr-1 Dcwr-2 fusion-compatible and the Dcwr-1 Dcwr-2 fusion-incompatible pools of progeny (Fig. 1B). Upon further inspection of this 3 Mb region, mapped reads coverage to NCU05915 were significantly lower in Dcwr-1 Dcwr-2 fusion-incompatible progeny pools compared to Dcwr-1 Dcwr-2 fusion-compatible progeny pools (Fig. S1A).
Using assembled genome sequences of 23 N. crassa isolates (26), we analyzed polymorphisms at NCU05915 and linked loci (NCU05914, NCU05916, and NCU05917) (Fig.  S2). Among the 23 strains, alleles at NCU05914 and NCU05917 were highly conserved (.90 amino acid identity) (Fig. 1C, Fig. S1B and S2). In contrast, alleles of NCU05916 showed high sequence diversity and alleles fell into two haplogroups among the 23 wild isolates (Fig. 1C, Fig. S1B and S2). We defined the alleles of NCU05916 with high conservation to FGSC2489 (the laboratory strain; amino acid identity . 96%) as haplogroup I, and alleles that were highly similar to each other but different from haplogroup I alleles, and which included JW224, as haplogroup II (Fig. 1C, Fig. S1 and S2). Interestingly, all the strains within haplogroup II lacked the linked locus NCU05915, while within haplogroup I strains, NCU05915 alleles were highly conserved with above 98% amino acid identity (Fig. 1C, Fig. S1 and S2).
Cell fusion deficient phenotype of Dgt69-2 is suppressed by mutations in rfw-1. To determine whether gt69-2 and/or rfw-1 was responsible for cell fusion arrest, we generated Dgt69-2 and Drfw-1 single deletion mutants, and a Drfw-1Dgt69-2 double deletion mutant by replacing gt69-2, rfw-1, or the whole region containing both rfw-1 and gt69-2 with a hygromycin B-resistance cassette in an FGSC2489 background (see the Materials and Methods) ( Fig. S3A and B). Cell fusion assays were performed by pairing FM4-64-stained mutant germlings with FGSC2489 germlings expressing cytoplasmic green fluorescent protein (GFP). The Dgt69-2 and Drfw-1Dgt69-2 germlings underwent chemotropic interactions, but failed to complete cell fusion and cytoplasmic mixing with FGSC2489 germlings (Fig. 2B). In contrast, the Drfw-1 mutant showed a wild-type cell fusion phenotype when paired with FGSC2489. These data indicated that gt69-2 was required for successful cell fusion with its wild-type parental strain.
To confirm that a deletion of rfw-1 suppresses the cell fusion defect of Dgt69-2, we generated a second double mutant by introducing a Drfw-1 deletion into a Dgt69-2 mutant by replacing rfw-1 with a nourseothricin-resistance cassette (see the Materials and Methods) ( Fig. S3A and B). This independently derived double mutant (DNCU05915 Dgt69-2) (Table  S1) showed an identical slant phenotype to the Drfw-1 Dgt69-2 mutant (Fig. S3C) and, identical to the Drfw-1Dgt69-2 mutant, underwent fusion in self pairings but not when paired with FGSC2489 (Fig. S3D). These data supported the original observation that deletion of rfw-1 suppressed the cell fusion defects of the Dgt69-2 mutant.
To quantify cell fusion frequencies in the mutants relative to wild-type cells, we utilized a flow cytometry method based on a robust postfusion death response in germinated spores that is mediated by genetic differences at sec-9 (29, 30). In brief,  Table S1) paired with FM4-64-stained FGSC2489 (the parent of the Dcwr-1Dcwr-2 mutant) or Dcwr-1Dcwr-2 (GFP) germlings blocked in cell fusion when paired with FM4-64-stained wild isolate JW224 by epifluorescence microscopy. (B) SNP segregation on linkage group VI (from 1.2 Mb to 4.2 Mb) after bulk segregant analysis and sequencing of two pools of genomic DNA from FGSC2489 fusion-compatible versus fusion-incompatible progeny from a cross between FGSC2489 and JW224. Blue line: SNP frequency in pooled segregants compatible with FGSC2489. Red line: SNP frequencies in pooled segregants incompatible with FGSC2489. Black box shows the region of centromere. Red arrow shows the position of gt69-2 and rfw-1. (C) Genomic organization of gt69-2 (NCU05916) linked loci in FGSC2489 and wild isolates. The percentage identity of the predicted protein sequences from sequenced wild isolates was calculated using FGSC2489 as the reference. The strains lacking NCU05915 (rfw-1) are marked with a dash.
Li et al. Regulation of Cell Fusion in Neurospora crassa ® FGSC2489 and mutant strains were engineered to carry sec-9 GRD2 at the native sec-9 locus. When germlings carrying incompatible sec-9 alleles undergo cell fusion, cell death is induced within 20 min, which can be used as a proxy for cell fusion frequency using vital dyes and flow cytometry (29,30). FGSC2489 1 FGSC2489 sec-9swap pairings were used as a positive control and showed a high death rate (;22%), while a negative-control pairing between cells unable to complete cell fusion (FGSC2489 with cwr-1 JW228 1 FGSC2489 sec-9swap ) showed a low death frequency (;5%) (Fig. 2D), a value consistent with that previously reported (29). As predicted by microscopic analyses, the Dgt69-2 1 FGSC2489 sec-9swap pairings, the Dgt69-2 1 Dgt69-2 sec-9swap pairings, and the Drfw-1Dgt69-2 1 FGSC2489 sec-9swap pairings all showed a low death frequency (2 to 5%) (Fig. 2D), consistent with a block in cell fusion. In line with the microscopy results, the Drfw-1 1 FGSC2489 sec-9swap pairings and the Drfw-1 1 Drfw-1 sec-9swap pairings both showed a high level of death frequency, showing that cells lacking rfw-1 are not affected in cell fusion (Fig. 2D). The Drfw-1Dgt69-2 1 Drfw-1Dgt69-2 sec-9swap self-pairings also showed a high death frequency (Fig. 2D), confirming that the lack of rfw-1 suppressed the cell fusion defect of the Dgt69-2 mutant. Additionally, these data also showed that neither GT69-2 nor RFW-1 was essential for cell fusion, as Drfw-1Dgt69-2 germlings showed self-fusion frequencies that were slightly higher than parental WT germlings (Fig. 2D).
Genetic interactions between gt69-2 and rfw-1. The Dgt69-2 mutant showed a lower height of aerial hyphae compared to FGSC2489 (Fig. 3A), a phenotype that has been observed in other cell fusion mutants (21,32,33). However, this phenotype was not observed in the Drfw-1 or Drfw-1Dgt69-2 mutant strains, indicating that, analogously to the cell fusion process, the short aerial hyphae phenotype of Dgt69-2 was suppressed by deletion of rfw-1. To test whether the Dgt69-2 mutant showed a lower growth rate, we inoculated hyphal plugs or conidial suspensions of each strain on Vogel's minimal medium (VMM) agar plates and measured the diameters of colonies up to 2 days postinoculation. When a conidial suspension was inoculated onto plates, the Dgt69-2 mutant showed a smaller colony diameter and fewer aerial hyphae compared to FGSC2489 ( Fig. 3B and C). By plotting colony diameter over time, the Dgt69-2 showed a lower growth rate for 24 h, consistent with a lag in colony establishment, a phenotype that has also been observed in other cell fusion mutants (21) (Fig. 3C). In contrast, with hyphal plug inoculations-that is, after the colony was already established-the Dgt69-2 mutant and FGSC2489 showed a similar growth rate (Fig. 3C). These data indicated that gt69-2 was dispensable for growth rate of a mycelial colony, but important for colony establishment via germling fusion.
Cells lacking gt69-2 affect oscillation of MAK-2 and are blocked in cell wall dissolution. To assess when the cell fusion defect occurred in Dgt69-2 cells, we first used transmission electron microscopy to determine whether the fusion defect in Dgt69-2 cells was due to a failure in cell wall dissolution or in membrane merger. In FGSC2489 1 FGSC2489 samples, cell wall and plasma membrane dissolution at the point of contact between germling fusion pairs was easily observed (Fig. 5A). In contrast, in Dgt69-2 1 Dgt69-2 pairings, we failed to find cell wall dissolution at contact points ( Fig. 5A), and accumulation of cell wall material at cell-cell contact sites was not observed, in contrast to cell pairings between incompatible cwr strains (29). These data indicated that the block of cell fusion in Dgt69-2 mutant was caused by failure of cell wall breakdown upon contact between Dgt69-2 cells.
During chemotropic interactions between compatible cells, the mitogen-activated protein kinase (MAPK) signal transduction protein complex (NRC-1, MEK-2, MAK-2, and the scaffold protein HAM-5) are recruited to conidial anastomosis tubes (CATs) (19). The MAK-2 complex assembles and disassembles at CAT tips every 8 to 10 min; chemical inhibition of the phosphorylation activity of MAK-2 results in immediate cessation of chemotropic growth (20). A second protein complex bearing SOFT (SO) also assembles and disassembles at CAT tips, but perfectly out of phase with the MAK-2 complex (20). FGSC2489 (MAK-2-GFP) 1 FGSC2489 (SOFT-dsRED) cells display oscillation of MAK-2 and SOFT to CATs during chemotropic interactions until physical contact. Previously, we showed that in cell pairings between incompatible cwr strains, MAK-2 and SO continued to oscillate at the contact point, consistent with an inability of cwr incompatible cells to transit from chemotropic growth to cell wall dissolution (29).
To further explore the block in self cell fusion in the Dgt69-2 cells, we analyzed MAK-2-GFP localization in Drfw-1(mak-2-gfp) germlings, in Dgt69-2 (mak-2-gfp) germlings, and in Drfw-1Dgt69-2(mak-2-gfp) germlings. In wild-type pairings, MAK-2-GFP shows dynamic localization to CATs during chemotropic interactions, localizing to one CAT tip while disappearing from its partner cell every ;4.5 min (Fig. 5B). Consistent with microscopic observations showing wild-type levels of cell fusion, the Drfw-1 cells showed normal dynamics of MAK-2 oscillation during chemotropic interactions Regulation of Cell Fusion in Neurospora crassa ® (Fig. 5C). In pairings between Dgt69-2 cells, oscillation of MAK-2 was observed during chemotropic interactions, but when Dgt69-2 germlings were in close proximity, MAK-2 localization to CATs was no longer observed (Fig. 5D). Additionally, MAK-2 localization at the contact point between Dgt69-2 germlings was not observed, which is apparent in wild-type pairings. These data indicated that Dgt69-2 germlings were affected during interactions when cells were in close proximity and in subsequent cell wall dissolution. Importantly, normal MAK-2-GFP dynamics during chemotropic interactions were restored in self pairings of Drfw-1Dgt69-2 germlings, consistent with the suppression of the cell fusion defect of the Dgt69-2 cells by deletion of rfw-1 (Fig. 5E).
GT69-2 and RFW-1 localization, overexpression phenotypes, and sensitivity to cell wall stress. Both GT69-2 and RFW-1 have predicted signal peptides. To characterize the subcellular localization of GT69-2 and RFW-1, we fused GFP to the N-terminal region of the predicted proteins immediately after the predicted signal peptides. The GFP-fused gt69-2 and rfw-1 were driven by the ccg-1 promoter and expressed in Dgt69-2 and Drfw-1 cells, respectively; GFP fluorescence was not observed in constructs using the gt69-2 or rfw-1 native promoters. The ccg-1-regulated gfp-gt69-2 construct fully complemented the growth and cell fusion defects of the Dgt69-2 mutant (Fig.  S3E). Both GFP-GT69-2 and GFP-RFW-1 showed a similar subcellular localization pattern as numerous fluorescent punctate structures in hyphal compartments (Fig. 6A and B), with a similar localization pattern in germlings (Fig. S4). It is likely that increased protein levels from ccg-1-driven gt69-2 and rfw-1 expression resulted in a more abundant localization to Golgi. Localization of GFP-GT69-2 or GFP-RFW-1 to puncta within the cell did not change in germlings undergoing chemotropic interactions or cell fusion. To determine which organelles the puncta were, we coexpressed GFP-GT69-2 or GFP-RFW-1 with the Golgi marker mCherry-VPS-52 or the ER marker mCherry-ERV-25 in heterokaryotic strains. Colocalization of GFP-GT69-2 or GFP-RFW-1 with the ER marker ERV-25 was not observed, however, many of the GFP-GT69-2 and GFP-RFW-1 puncta colocalized with mCherry-VPS-52 ( Fig. 6A and B). These data suggested that the punctate structures to which GFP-GT69-2 and GFP-RFW-1 localized were Golgi compartments.
The gt69-2 locus encodes an alpha-1,3-mannosyltransferase predicted to transfer a mannosyl group to either a carbohydrate or a lipid. We therefore hypothesized that loss of gt69-2 might affect aspects of the cell wall biosynthesis. To test this hypothesis, we assessed growth of Drfw-1, Dgt69-2, and Drfw-1Dgt69-2 mutants on agar media containing different cell wall stress drugs, including the b-1,3-glucan synthase inhibitor caspofungin and two different anionic dyes that bind chitin and block chitin-glucan cross-linking, calcofluor white and Congo red. Similar to the parental strain FGSC2489, the Drfw-1 and Drfw-1Dgt69-2 mutants were mildly sensitive to all three drugs (Fig. S5). Consistent with conidial inoculations, the Dgt69-2 mutant showed a slight growth defect in drug-free medium. However, these defects were not exacerbated on caspofungin, calcofluor white, or Congo red, indicating that the absence of gt69-2 did not result in major cell wall defects.
Alleles at gt69-2 and rfw-1 show evidence of balancing selection. Genes that regulate allorecognition, such as the major histocompatibility complex (MHC) in humans, the S locus in plants, allorecognition loci in colonial ascidians, and heterokaryon incompatibility loci in fungi, often show evidence of balancing selection, which includes the presence of discrete haplotypes in populations, nearly equal frequency of allelic classes in population samples, and transspecies polymorphisms (26,(34)(35)(36). In N. crassa populations, gt69-2 alleles fell into two discrete haplotypes, suggesting a role in allorecognition ( Fig. 2A). In strains containing rfw-1, the gene was always linked with gt69-2 and was highly conserved among isolates. Phylogenetic trees were constructed to test whether allelic polymorphisms at rfw-1 (NCU05915) and gt69-2 (NCU05916) were retained among different Neurospora species. Consistent with their potential role in allorecognition, the gt69-2 alleles clustered by haplogroup rather than by species (Fig. 7B). The gt69-2 alleles from Neurospora discreta and Neurospora tetrasperma isolates grouped into the same two N. crassa haplogroups. Similar to N. crassa, the haplogroup I gt69-2 alleles in both N. discreta and N. tetrasperma were linked to rfw-1, while species of all strains within haplogroup II lacked rfw-1. The transspecies polymorphisms observed in the gt69-2 alleles suggested that this locus was under balancing selection and that allelic polymorphisms at this locus predates divergence of these species. We tested this hypothesis by calculating the Tajima's D values for the gt69-2 alleles. The high, positive, and significant Tajima's D values calculated for gt69-2 (Tajima's D = 2.07708; P , 0.05), but not NCU05914 (Tajima's D = 0.73738; P . 0.1) or NCU05917 (Tajima's D = 1.07540; P . 0.1), indicated that gt69-2 is under balancing selection in Neurospora species.
To assess whether allelic polymorphisms were present in other species of fungi, we analyzed the gt69-2 and rfw-1 homologs among various species of Fusarium, in particular, Fusarium oxysporum, as genome sequences for multiple isolates are available (Table S3). In Fusarium species, most strains have more than one paralog of gt69-2 and rfw-1 (Fig. S6). However, in strains of different species of Fusarium, if rfw-1 was present, it was always linked with gt69-2, although gt69-2 loci were identified that lacked linked rfw-1. In a sample of F. oxsporum isolates, although variation was observed in the number of gt69-2 and rfw-1 homologs in these isolates, allelic polymorphisms and discrete haplotypes were not observed (Fig. S6B).

DISCUSSION
In this study, we identified a linked gene pair, gt69-2 and rfw-1, that functions to regulate somatic cell fusion in N. crassa. The gt69-2 locus is predicted to encode a CAP59-like a-1,3-mannosyltransferase and, based on its similarity to C. neoformans CMT1, to catalyze the transfer of mannose from GDP-mannose to a-1,3-linked mannose disaccharides (31). A paralog of CMT1 in C. neoformans, CAP59, is required for capsule synthesis by playing a role in the export of the capsular polysaccharide glucuronoxylomannan (31). Both gt69-2 and CAP59 orthologs belong to glycosyltransferase family 69 and contain the conserved CAP59 family alpha-1,3-mannosyltransferase catalytic domain. In Aspergillus fumigatus, the Golgi-localized protein ClpA adds an alpha-1-3linked mannose to glycosylphosphatidylinositol (GPI) anchors (37); clpA is a homolog of Cap59. GPI anchors are important for anchoring cell surface proteins to the plasma membrane/cell wall (38). The attachment of the GPI anchor occurs in the ER, but the understanding of the maturation of the GPI anchor that occurs in the Golgi is limited.
We hypothesized that GT69-2 functions to modify secreted protein(s), such as GPI-anchored proteins, destined for the cell wall or plasma membrane, or that a small fraction of GT69-2 is trafficked to the cell surface during chemotropic interactions, modifying proteins important for late stages of MAK-2 signaling and cell wall remodeling/dissolution during the process of cell fusion. A wrinkle in this hypothesis was the observation that loss-of-function mutations in rfw-1 suppressed the cell fusion defect of the Dgt69-2 mutant; Dgt69-2Drfw-1 mutants were fusion competent. These data indicated that neither GT69-2 nor RFW-1 are essential for cell fusion in N. crassa, but rather, in the absence of GT69-2, RFW-1 functions to block cell fusion. We predict that in the absence of GT69-2, RFW-1 may inappropriately modify a protein or block secretion of a protein needed for mediating the transition from chemotropic interactions to cell wall dissolution, resulting in the loss of MAK-2 localization at cell contact sites and cessation of the cell fusion process. Localization of MAK-2 to the fusion pore as cell wall dissolution and membrane merger are occurring has been reported previously (20), and MAK-2 kinase activity is required for cell wall dissolution (39).
Consistent with the above hypothesis, overexpression of rfw-1 resulted in a block in cell fusion, even in the presence of gt69-2. The overexpression rfw-1 strain also showed a conidial separation deficiency associated with an inability to remove cell wall material at the double-doublet stage of conidial development. The phenotype of the rfw-1 overexpression strain most closely resembles the csp-2 mutant in N. crassa, where csp-2 encodes a homolog of grainy head-like transcription factors (40). An inability to remove the thin connectives between adjacent conidia has been associated with a decrease in autocatalytic activity of the cell wall, hypothesized to be due to a lack of secreted enzymes, such as chitinases (41); a gene encoding a chitinase and additional proteins associated with cell wall structure were identified as transcriptional targets of CSP-2 (40). Two cell wall glycosyl hydrolases, the CGL-1 b-1,3-glucanase and the NAG-1 exochitinase, function in remodeling the cell wall between adjacent conidia to facilitate conidia formation and dissemination (42). Two additional predicted GPI-anchored proteins, BGT-1 and BGT-2, encoding predicted b-1-3 endoglucanases (GH17 family) (43), localize to double-doublets in developing conidia and also to fusion points of germlings and hyphae (44). The Dbgt-1 and Dbgt-2 mutants display a deficiency in conidial separation, but do not display a cell fusion defect (44). Other mutants in N. crassa that show defects in conidial separation do show defects in cell fusion, however, including loss-of-function mutations in whi-2, csp-6, and amph-1 (23,32). CSP-6 and WHI-2 physically interact (45) and WHI-2, which localizes to the cell periphery, is required for signaling during chemotropic interactions via the MAK-2 MAPK pathway (23). Future studies to identify targets of RFW-1 and GT69-2 should help to understand the molecular basis of the cell wall remodeling process regulated by the RFW-1/GT69-2 system.
In the genomes of Fusarium and Neurospora species, all predicted rfw-1 genes were always linked to gt69-2 genes, although homologs of gt69-2 occurred without a linked rfw-1 gene (Fig. S6). These observations suggest that GT69-2 and RFW-1 also function as a pair in species other than in N. crassa. Coevolution of linked genes to maintain physical or functional interactions of their products occurs via coordinated sequence changes between the gene pairs (46). In Neurospora species, gt69-2 orthologs found in two haplogroups showed evidence of balancing selection, similar to other systems regulating allorecognition (25,27,29,30,47). However, expression of a gt69-2 JW224 (haplogroup II allele) in a gt69-2 FGSC2489 (haplogroup I allele) strain was insufficient to activate allorecognition and block cell fusion. The gt69-2 JW224 allele was fully functional, as it fully complemented the fusion-deficiency phenotype of a Dgt69-2 mutant. One possible explanation is that the gt69-2 alleles from haplogroup II have adapted to the loss of rfw-1, while haplogroup I strains need both gt69-2 and rfw-1 to correctly modify their targets in the Golgi. Alternatively, it is possible that the evolutionary forces driving balancing selection at gt69-2/rfw-1 do not reflect the function of these two proteins in cell fusion/conidial separation. Further work to identify the targets of the GT69-2/RFW-1 pair from haplogroup I relative to GT69-2 from haplogroup II will help to resolve this question, in addition to identifying cell membrane/cell wall-associated proteins required for late functions of MAK-2 signaling involved in cell wall dissolution and membrane merger during somatic cell fusion.

MATERIALS AND METHODS
Strains and growth conditions. Standard procedures and protocols for N. crassa can be found on the Neurospora homepage at the Fungal Genetics Stock Center (FGSC, www.fgsc.net/Neurospora/ NeurosporaProtocolGuide.htm). Vogel's minimal medium (VMM) (with supplements, if required) was Li et al.
used to culture all strains (48). Crosses were performed on Westergaard's synthetic crossing medium (49). All the strains used in this study are listed in Table S1 in the supplemental material. The wild N. crassa isolates from a Louisiana population have been previously described (25,26,50). FGSC2489 served as the wild-type (WT) control for all experiments and the parental strain for gene engineering, unless stated otherwise.
Strain construction. All gene deletion constructs were generated by double-joint PCR (25,51). The deletion mutants were obtained as described (25,29). For the Drfw-1Dgt69-2 double mutant, the whole region containing both NCU05915 and NCU05916 was replaced with the hygromycin B-resistance cassette in FGSC2489. For the independently derived DNCU05915 Dgt69-2 double mutant, rfw-1 was replaced with the nourseothricin-resistance cassette (52) in the Dgt69-2 mutant. Putative deletion mutants were screened for drug resistance and further confirmed by PCR (Fig. S3A and B). The primers are listed in Table S2.
The FGSC2489 sec-9swap strain, which was engineered to carry sec-9 GRD2 at the native sec-9 locus, has been previously described (30). The Drfw-1 and/or Dgt69-2 mutants were crossed with FGSC2489 sec-9swap to obtain the resulting sec-9swap strains.
Bulk segregant analysis. Bulk segregant analysis (BSA) followed by whole-genome resequencing was performed as previously described (25). Approximately 60 ng of genomic DNA from ;49 progeny strains in each DNA pool was used for library preparation and sequencing. All paired-end libraries were sequenced on a HiSeq2000 sequencing platform using standard Illumina operating procedures (QB3 Genomics Lab, University of California, Berkeley).
Microscopy. Cell fusion experiments were performed as described (25). Cytoplasmic or histone 1tagged GFP-expressing cells and FM-64-stained (Thermo Fisher Scientific) cells were mixed in a 1:1 proportion and incubated on VMM plates at 30°C in the dark for 4 h. Cytoplasmic mixing was examined with a Zeiss Axioskop 2 microscope equipped with a Q Imaging Retiga-2000R camera (Surrey) using a 40Â/1.30 Plan-Neofluar oil immersion objective and the iVision Mac 4.5 software.
Heterokaryotic strains bearing both GFP and mCherry fluorescent proteins were prepared as described (25) for colocalization analysis. Images were taken with a Leica SD6000 confocal microscope equipped with a Yokogawa CSU-X1 spinning disk head, and a 488-nm or 561-nm laser controlled by Metamorph software.
For MAK-2 oscillation experiments, conidia from strains expressing MAK-2-GFP were prepared for microscopy as described (25). Time-lapse microscopy was performed using the confocal microscope system as described above. Images were captured at 30 s intervals. The software ImageJ was used for image processing. Fluorescence signals were quantified as previously described (20).
Transmission electron microscopy. Conidia were inoculated in 100 ml of liquid VMM at a final concentration of 10 6 conidia/ml for 5 hat 30°C (shaking at 220 rpm for 2.5 h and standing for 2.5 h). Cells were harvested by centrifugation and then fixed with electron microscopy fix buffer (2% glutaraldehyde, 4% paraformaldehyde, 0.04 M phosphate buffer [pH 7.0]), followed by 2% KMnO 4 treatment. Samples were dehydrated using a graded ethanol series before embedding the samples in resin.
Flow cytometry. Flow cytometry was performed as described (29). For each experiment, 20,000 events per sample were recorded on a BD LSR Fortessa X-20 (BD Biosciences, Franklin Lakes, NJ, USA). Cell death frequencies were analyzed with a specifically designed MATLAB script (29). Each experiment was performed at least three times.
Growth assays. To evaluate growth rate, a hyphal plug (1 mm 2 ) or 5ml of a conidial suspension (10 6 conidia/ml) was inoculated onto the center of 14.2-cm diameter petri dishes and grown at 30°C in constant dark. The colony diameter was recorded twice a day.
Cell wall stress assays were conducted on VMM 1 FGS with 1.3 mg/ml caspofungin, 1.5 mg/ml calcofluor white, or 1 mg/ml Congo red as described (55). A 1:5 dilution series was prepared starting with a concentration of 10 6 conidia/ml. Conidial solutions were then spotted onto freshly poured plates at 5 ml per spot.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.