Identification of Communication Groups in N. crassa Populations

Kind discrimination was neglected in previous studies on germling communication in N. crassa, as strains were used whose genetic background was identical to the commonly used laboratory strain (FGSC 2489) [43,44,50]. To assess whether germlings of different genetic backgrounds can undergo productive chemotropic interactions, we took advantage of a N. crassa population isolated from Louisiana, United States; the laboratory strain (FGSC 2489) is a member of this population [51]. RNAseq data showed a substantial level of polymorphism (on order of two single nucleotide polymorphisms [SNPs] per kbp), while analyses of population structure revealed no subdivision [51–53]. We randomly picked 14 isolates from this population and analyzed chemotropic interactions (defined as reoriented growth of germlings toward each other; Fig 1A) between genetically identical germlings from the same isolate (self-communication) versus chemotropic interactions between germlings from the 14 isolates and FGSC 2489 (non-self-communication) using differential fluorescence labeling (see Materials and Methods). Self-communication frequencies among the wild isolates varied between approximately 50% and 95% (Figs 1A, 1B and S1). However, when germlings of these 14 wild isolates were paired with FGSC 2489 (non-self-communication), some pairings showed very low communication frequencies, while others showed communication frequencies with FGSC 2489 that were similar to self-communication frequencies (Figs 1A, 1B and S1). Importantly, this communication phenotype was not linked to the mating type of the strains (S1 Table). We therefore assessed self and non-self germling communication phenotype of the remaining members of the Louisiana population (95 strains) (see Materials and Methods; S1 Table). From these analyses, three communication groups were defined. While genetically identical and non-identical germlings within a communication group showed robust chemotropic interactions, germlings from different communication groups, even when in close proximity, grew past each other to find a germling of their own communication group (S1 and S2 Movies). The first communication group (CG1) contained 29 strains, which showed similar communication frequencies between and within strains and which included FGSC 2489 (Fig 1C, orange). The second communication group (CG2) contained 51 strains (Fig 1C, green), while the third communication group (CG3) contained 21 strains (Fig 1C, blue; S1 Table). The remaining nine strains (Fig 1C, black) did not produce sufficient asexual spores to determine CG affiliation. These observations indicated that the germling communication trait in N. crassa functioned in assortative kind recognition and occurs at a distance.

To determine whether communication groups are unique to the Louisiana population, we used tester strains for each communication group (FGSC 2489, CG1; JW262, CG2; P4483, CG3; framed in Fig 1C) and evaluated communication frequencies with other N. crassa population samples (isolates from Haiti, Panama, Costa Rica, Puerto Rico, Texas, Florida, Venezuela, Guyana). All of the wild isolates from these different N. crassa populations communicated with one of the three Louisiana communication group tester strains (CG1, CG2, or CG3; S1 Table). Thus, communication groups were not unique to the Louisiana population, but also occurred in other wild populations of N. crassa.

Communication Group Affiliation Is Associated with Genomic Rearrangements and with Loci that Encode Highly Divergent Alleles

Based on the distribution of communication groups in the Louisiana population, we reasoned that genes that conferred kind recognition in N. crassa functioned as a Mendelian trait. To test this hypothesis, we used crossings to determine the number of loci mediating CG affiliation, making use of the fact that the affiliation of strains in the different CGs does not affect sexual compatibility [54]. We crossed a CG1 strain (FGSC 2489) with a CG2 strain (JW258), a CG1 strain (FGSC 2489) with a CG3 strain (D113), and a CG2 strain (JW242) with a CG3 strain (D113). In all crosses, the CG phenotype of the progeny segregated approximately 1:1, with approximately one-half of the progeny communicating with one parental strain and the second half of the progeny communicating with the other parent, consistent with our prediction that a single locus or closely linked loci were involved in kind recognition and determined CG affiliation. To identify the CG locus, we performed a bulk segregant analysis followed by whole genome resequencing of progeny from a cross between a CG1 strain (FGSC 2489) and a CG2 strain (JW258). Genomic DNA from 46 CG1 progeny or 46 CG2 progeny was isolated, pooled, and sequenced, revealing a ~100 kbp region on the right arm of linkage group V that showed segregation of SNPs between CG1 versus CG2 progeny at ~100% frequency, which was embedded within a larger divergent region of ~450 kbp (Fig 2A). A random SNP distribution of ~50% was observed for the remaining six linkage groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Communication group haplotypes.

(A) SNP segregation on linkage group V after bulk segregant analyses and sequencing using pooled genomic DNA from 46 CG1 versus 46 CG2 progeny. Blue line: SNP frequency in pooled segregants that communicated with the CG1 tester. Red line: SNP frequencies in the pooled segregants that communicated with the CG2 tester. SNP frequencies are shown in comparison to the reference genome (FGSC 2489) [56]. (B) Genomic organization of communication group haplotypes (CGH) in wild isolates. Genomic rearrangements within the CGHs spanned the genetic interval between NCU07191 and NCU07194 and included duplications of NCU07192 (doc-2, doc-3), a deletion of NCU07192, and inversions. Alleles at NCU07191 (doc-1) and NCU07192 (doc-2, doc-3) within a CGH show high DNA sequence identity, but are polymorphic between CGHs. The percent DNA identity between alleles in members of the different CGH groups across the genetic interval in comparison to FGSC 2489 (a member of CGH1) are shown.

https://doi.org/10.1371/journal.pbio.1002431.g002

We used resequencing data from 26 wild isolates from the Louisiana population [55] to define allelic sequences at the CG locus within the ~100 kbp region. Among these 26 strains, seven isolates were members of CG1, 15 isolates were in CG2, and three isolates were members of CG3, while one isolate did not produce sufficient asexual spores to determine CG affiliation (asterisks in Fig 1C; S1 Table). Analysis of the ~100 kbp interval in the genomes of these 26 isolates revealed a 14 kbp region that showed five different genomic rearrangements that spanned four loci, NCU07191 to NCU07194 (gene nomenclature based on the reference genome from FGSC 2489 [56]), referred to as communication group haplotype 1 through 5 (CGH1–5) (Fig 2B). Of the 26 isolates, seven showed a CGH1 organization, five showed a CGH2 organization, three had a CGH3 organization, four had a CGH4 organization, and seven had a CGH5 organization. The CGH2 and CGH4 strains contained a duplication of NCU07192 (doc-3), while CGH5 strains did not contain NCU07192 (Fig 2B). Inversions of NCU07192 (CGH2 and CGH3 isolates) or the entire genetic interval between NCU07191 and NCU07194 were also observed (CGH4 and CGH5 isolates). Of the genes within this region, NCU07191 and NCU07192 showed ~43% DNA sequence identity, suggesting that they are paralogous. Paralogy was also supported by analyses using OrthoMCL (OG5_241519) [57].

To determine whether structural differences between CGHs were also associated with nucleotide differences, we used sequence alignments to characterize the nature and level of variability at genes within the haplogroups. Among five loci in the genetic interval associated with CGH, NCU07190, NCU07193, and NCU07194 displayed a high level of conservation among all 26 isolates (>90% DNA sequence identity with few nucleotide substitutions; Fig 2B; S2 Table). In contrast, NCU07191 and NCU07192 displayed high levels of allelic variability among the 26 isolates, with alleles falling into five main groups, which correlated with the genomic rearrangements among the five CGHs. Alleles at NCU07191 and NCU07192 showed only ~50% DNA sequence identity, with members of the different CGHs being highly divergent (0.22 to 0.74 differences/bp between NCU07191 alleles and 0.26 to 0.87 differences/bp between NCU07192 alleles; S2 Table). The predicted proteins encoded by NCU07191 and NCU07192 among members of the different CGH groups were also highly variable, with only ~35% amino acid identity (CGH1 versus all other CGH isolates), with few regions showing high conservation in all of the predicted NCU07191 or NCU07192 proteins (S2 Fig). In contrast, isolates within a single CGH showed DNA and amino acid identity at NCU07191 and NCU07192 that were comparable with the rest of the genome (up to 99% DNA and over 95% amino acid sequence identity) (S2 Fig), with the exception of the CGH1 isolates. Within the CGH1 isolates, alleles at NCU07191 and NCU07192 fell into two different subgroups (CGH1A and CGH1B), with ~70% DNA and ~60% amino acid sequence identity between members of the two subgroups (S2 Table). DNA sequence alignments of the genetic interval between NCU07191 and NCU07194 of the CGH1 isolates indicated that there were CGH1A- and CGH1B-specific indels and SNPs in the intergenic region between NCU07190 and NCU07193 (S3 Fig). Similarly, both CGH-specific SNPs resulting in amino acid substitutions and CG-specific indels differentiated isolates between the different CGH groups (S2 Fig).

The presence of five genomic CGHs with only three phenotypically distinguishable communication groups within the Louisiana population prompted us to reevaluate the communication phenotype of the 26 sequenced strains. For members of CG1 and CG3, the germling communication phenotype was completely correlated with CGH (S1 Table; Fig 2B); no difference in germling communication frequency was observed in isolates between the subgroups CGH1A and CGH1B. Unlike the CG1/CGH1 and CG3/CGH3 strains, the 15 strains defined as CG2 displayed multiple different genomic arrangements in this region (CGH2, CGH4, and CGH5; Fig 2B). However, using an isolate from each defined communication group (FGSC 2489, CG1; JW262, CG2; P4483, CG3), we identified a fourth phenotypic communication group (CG4). Germlings within this communication group (D111, JW179, P4479, P4489; S1 Table) showed very low communication frequencies with CG1 germlings, but underwent robust chemotropic interactions with both CG2 and CG3 germlings (S1 Table). These CG4 strains all showed the same genomic organization in the NCU07191 to NCU07194 genetic interval (CGH4), and with high DNA and amino acid sequence identity (>99%). Thus, of the 26 sequenced strains, seven isolates fell into CG1, with a CGH1 genomic organization; 11 isolates fell into CG2, with a CGH2 or CGH5 genomic organization; three isolates fell into CG3, with a CGH3 genomic organization; and four isolates were CG4 with a CGH4 genomic organization. An additional isolate (JW246) did not produce asexual spores, but had a CGH2 genomic organization.

We refer to NCU07191, NCU07192, and the duplicated version of NCU07192 as determinant of communication 1, 2, and 3 (doc-1, doc-2, and doc-3, respectively). doc-1, doc-2, and doc-3 encode predicted hypothetical proteins (DOC-1, 828 aa; DOC-2, 839 aa; DOC-3, 920 aa). All of the DOC-2 proteins from the different CGH groups have a predicted OmpH-like outer membrane protein domain, although with some variability in conservation (S2 Fig), while DOC-1 and DOC-3 lack any identifiable functional domains. DOC-1 and DOC-2 contain one or two predicted transmembrane domains, while no transmembrane domain was predicted for DOC-3. Conserved homologs of doc-1, doc-2, and doc-3 were identifiable by BLAST in the Sordariales (order within the class Sordariomycetes in the division Ascomycota), but were not obvious in more distantly related fungal species.

doc-1 and doc-2 Determine Communication Group Affiliation

The association of communication group phenotype with CGH supported the hypothesis that the doc genes confer communication group specificity. To evaluate this hypothesis, we examined strains carrying deletions of doc-1 or doc-2 for communication phenotype. Strains carrying deletions of doc-1 (Δdoc-1) or doc-2 (Δdoc-2) in the CG1 background (FGSC 2489) [58] were macroscopically indistinguishable from FGSC 2489. To determine germling communication group phenotype in these deletion strains, we constructed Δdoc-1 and Δdoc-2 strains carrying a gene encoding cytoplasmic green fluorescent protein (GFP). Conidia of the communication group tester strains (FGSC 2489, JW262, or P4483) were stained with the membrane-selective endocytic dye FM4-64, mixed with the Δdoc-1 (GFP) or Δdoc-2 (GFP) strains, and subsequently analyzed for germling communication frequencies (Fig 3A). These analyses revealed that both the Δdoc-1 and Δdoc-2 germlings were impaired in self-communication: communication between isogenic Δdoc-1 germlings was reduced to 48 ± 14%, as compared to the parental strain frequency of 84 ± 7%, while self-communication frequency in Δdoc-2 germlings was reduced to 37 ± 16%.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Communication interactions of Δdoc-1, Δdoc-2, Δdoc-1 Δdoc-2, and Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings.

(A) Self-communication and non-self-communication frequencies between germlings of different mutants and the communication group tester strains. One-color bars denote self-communication frequencies between genetically identical germlings from a single strain, while two-color bars denote communication frequencies between genetically different germlings. Δdoc1Δdoc2^CG3 denotes Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3). CG tester strains were FGSC 2489, CG1; JW262, CG2; P4483, CG3. (B) Self-communication frequencies of CGH5 germlings (JW220) and non-self-communication frequencies of CGH5 germlings with CG1 (FGSC 2489) germlings (CG1 + CGH5), which show very low communication frequency. In contrast, Δdoc-1 Δdoc-2 germlings (deletion strain derived from FGSC 2489) show high communication frequencies with a CGH5 strain (JW220) (Δdoc-1 Δdoc-2 + CGH5). (C) Self-communication frequencies of Δdoc-1 Δdoc-2 germlings (left grey bar), self-communication frequencies of Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings (right grey bar), and non-self-communication frequencies of Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings with Δdoc-1 Δdoc-2 germlings (right bar). Δdoc1Δdoc2^CG3 denotes Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3). Experiments for A, B, and C were performed in triplicate with at least 100 germling pairs counted for each experiment. Black bars indicate standard deviations (Student's t test, *: p < 0.05, **: p < 0.01; see S1 Data for numerical values). (D) Quantitative heterokaryon test, in which conidia of CG1 strains with complementary auxotrophic markers (FGSC 4564; ad-3B + FGSC 9716; his-3), or conidia from Δdoc-1 Δdoc-2 strains with complementary auxotrophic markers (Δdoc-1 Δdoc-2; his-3 + Δdoc-1 Δdoc-2; pyr-4) or conidia from a CG1 strain and a Δdoc-1 Δdoc-2 strain with complementary auxotrophic markers (FGSC 4564; ad-3B + Δdoc-1 Δdoc-2; his-3) were mixed and spread on modified VMM agar plates (see Materials and Methods for details). Colonies form only if germlings/hyphae containing complementary auxotrophic markers undergo fusion to form a prototrophic heterokaryon. Images show the results of one out of three experiments (see S4E Fig for quantitative data).

https://doi.org/10.1371/journal.pbio.1002431.g003

Although self-communication frequency was reduced in Δdoc-1 and Δdoc-2 germlings, non-self-communication frequencies with germlings from the different communication group tester strains were similar to each other and to the self-communication frequencies of each deletion strain. For example, when Δdoc-1 germlings were paired with either CG2 or CG3 germlings, non-self-communication frequencies were similar or even slightly higher than self-communication frequencies of Δdoc-1 germlings (65 ± 7% for Δdoc-1 + CG2 germlings; 71 ± 7% for Δdoc-1 + CG3 germlings; Fig 3A). Similarly, non-self-communication frequencies of the Δdoc-2 germlings with CG2 and CG3 tester strains were not significantly different from self-communication frequencies (Fig 3A), although communication frequencies of Δdoc-2 germlings with the parental CG1 strain were higher (48 ± 6%) than with the CG2 (18 ± 9%) or CG3 tester strains (12 ± 5%). These data indicate that doc-1 and doc-2 are essential for mediating communication group discrimination in N. crassa.

Germlings from a Δdoc-1 Δdoc-2 Mutant Switch Communication Group

Since both the Δdoc-1 and Δdoc-2 mutants showed reduced germling communication frequencies, but no significant difference in the frequency of communication with members of the three communication groups, we hypothesized that a Δdoc-1 Δdoc-2 double mutant would be completely deficient in communication and cell fusion. To test this hypothesis, we created a Δdoc-1 Δdoc-2 mutant by homologous recombination (see Materials and Methods). As with the Δdoc-1 or Δdoc-2 single mutants, the Δdoc-1 Δdoc-2 mutant was morphologically indistinguishable from its parental strain (FGSC 2489), a phenotype different than other fusion mutants, which display a “flat” phenotype [50]. To our surprise, unlike the single Δdoc-1 or Δdoc-2 mutants, the Δdoc-1 Δdoc-2 mutant showed a self-communication frequency that was indistinguishable from its parental strain (~85%; Fig 3A). Even more surprisingly, the communication frequency between Δdoc-1 Δdoc-2 germlings and their otherwise isogenic parental CG1 strain (FGSC 2489) was extremely low (~2%). The Δdoc-1 Δdoc-2 germlings also showed very low communication frequency with the CG3 tester strain (P4483, ~8%; Fig 3A). In contrast, the Δdoc-1 Δdoc-2 germlings communicated fairly well with the CG2 tester strain (JW262, ~50%), although still significantly less than self-communication frequencies of Δdoc-1 Δdoc-2 germlings.

The communication phenotype of the Δdoc-1 Δdoc-2 germlings suggested that DOC-1 /DOC-2 negatively regulate germling communication behavior and that removal of DOC-1/DOC-2 resulted in the generation of a new communication group. We further tested this hypothesis by evaluating the communication phenotype of Δdoc-1 Δdoc-2 germlings with additional members of CGH2, CGH3, CGH4, and CGH5. A CGH3 (D113) strain and CGH4 strains (D111, P4489) showed very low communication frequencies with Δdoc-1 Δdoc-2 germlings (~10%), while CGH2 strains (JW258, P4463) showed a reduction in communication frequency with Δdoc-1 Δdoc-2 germlings, as was observed for the CG2 tester strain JW262 (~50%; S4 Fig). However, in contrast to members of the other CGH groups, germling communication frequencies of the Δdoc-1 Δdoc-2 mutant with members of CGH5 (JW75, JW220, and JW242) were identical to self-communication frequencies (Fig 3B; S4 Fig). These data showed that the Δdoc-1 Δdoc-2 mutant behaved exactly like members of CGH5, which contained one copy of doc-1 but no copy of doc-2. Thus, the phenotype of the Δdoc-1 Δdoc-2 mutant with the CGH5 isolates defined a new communication group, termed CG5.

The availability of isogenic strains that only differed in communication behavior (FGSC 2489, CG1; Δdoc-1 Δdoc-2, CG5) allowed us to assess whether just the genetic difference at doc-1 and doc-2 was sufficient to affect the formation of heterokaryons (i.e., a syncytium of two or more genetically different nuclei), which is mediated by both germling and hyphal fusion. We introduced auxotrophic markers (his-3 or pyr-4) into a Δdoc-1 Δdoc-2 strain and first evaluated its ability to form (Δdoc-1 Δdoc-2; his-3 + Δdoc-1 Δdoc-2; pyr-4) heterokaryotic colonies as compared to strains isogenic to their parent (FGSC 2489) but that carry complementary auxotrophic markers (his-3 or ad-3B) (S4 Table). As shown in Fig 3D, heterokaryon formation was indistinguishable between (Δdoc-1 Δdoc-2; his-3 + Δdoc-1 Δdoc-2; pyr-4) strains and (his-3 + ad-3B) strains. In contrast, heterokaryon formation was drastically reduced when Δdoc-1 Δdoc-2; his-3 conidia were mixed with ad-3B conidia (Figs 3D and S4E). These data indicated that even if communication and cell fusion were essential for survival of N. crassa, differences in communication group affiliation almost completely prevented cooperation via heterokaryon formation.

doc-1 and doc-2 Are Sufficient for Communication Group Affiliation

To determine if doc-1 and doc-2 were sufficient for communication group affiliation, we cloned the doc-1 and doc-2 alleles from a CG3 strain (P4471; doc-1^CG3 and doc-2^CG3) and targeted them to the his-3 locus in the Δdoc-1 Δdoc-2 mutant. The Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) strain was macroscopically indistinguishable from the laboratory strain from which it was derived (FGSC 2489). Germlings from Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) showed high self-communication frequencies (~80%; Fig 3A and 3C). However, Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings showed greatly reduced communication frequency with their parental Δdoc-1 Δdoc-2 strain, showing that doc-1^CG3 and doc-2^CG3 were functional in this strain (Fig 3C). Germlings from the Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) strain also showed low communication frequency with the CG1 tester strain (FGSC 2489), but showed some communication with the CG2 tester strain (JW262), although germling communication frequencies were reduced (~50%; Fig 3A). However, the Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings communicated well with the CG3 tester strain (P4483) and the donor for doc-1^CG3 and doc-2^CG3 alleles (P4471) (Figs 3A and S4D). These data indicated that addition of doc-1^CG3 and doc-2^CG3 to the Δdoc-1 Δdoc-2 mutant was sufficient to switch communication group from CG5 to CG3. To confirm that the Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings no longer belonged to CG5, we tested communication frequencies of Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings with a CG5 strain (JW242). In contrast to Δdoc-1 Δdoc-2 germlings, Δdoc-1 Δdoc-2 (his-3::doc-1^CG3 doc-2^CG3) germlings showed low communication frequency with JW242 germlings (~15%; S4C Fig versus S4D Fig). Thus, swapping doc-1 and doc-2 alleles from a member of CGH1 with doc-1 and doc-2 alleles from a member of CGH3 was sufficient to switch communication group.

DOC-1 Oscillates to the Tips of Conidial Anastomosis Tubes and Fusion Hyphae during Chemotropic Interactions

Self-recognition between isogenic germlings requires the MAK-2 MAP kinase complex (NRC-1, MEK-2, MAK-2, and the scaffold protein HAM-5), which oscillates to the tips of conidial anastomosis tubes and completely out-of-phase with SOFT, a scaffold protein for the cell wall integrity MAPK pathway (Fig 1A) [45–47,59,60]. To determine the cellular location of DOC-1 and DOC-2 during germling communication, and its relationship with temporal patterns of signaling, we constructed strains bearing doc-1-gfp and doc-2-gfp alleles that were fully functional in restoring communication frequencies in Δdoc-1 or Δdoc-2 germlings, respectively (S4 and S5 Figs). In self pairings between Δdoc-1 (doc-1-gfp) germlings, DOC-1-GFP localized to intracellular punctae, which oscillated to the tips of conidial anastomosis tubes during chemotropic interactions (Fig 4A), with an interval of 8–10 min (S3 Movie; S5 Fig), an identical oscillation pattern to that of the MAK-2 complex and SOFT [44]. To investigate whether DOC-1 oscillates with MAK-2 or with SOFT during chemotropic interactions, we analyzed DOC-1-GFP oscillation in germlings undergoing chemotropic interactions with germlings bearing MAK-2-mCherry or SOFT-mCherry; DOC-1 oscillated with MAK-2, but completely out of phase to SOFT (Figs 5A, 5B and S5; S3 Movie). A heterokaryotic strain bearing both DOC-1-GFP and MAK-2-mCherry confirmed these observations (Fig 5C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Cellular localization of DOC-1-GFP and DOC-2-GFP.

(A) DOC-1-GFP showed dynamic localization to puncta at the tips of conidial anastomosis tubes during chemotropic interactions between genetically identical cells, with an oscillation period of 8–10 min within a single germling tip. See S5B Fig for oscillation intervals and S3 Movie for interacting germlings. (B) Oscillation of DOC-1-GFP in hyphae in a single colony undergoing chemotropic interactions prior to cell fusion. See Fig 5A for oscillation intervals and S4 Movie for interacting fusion hyphae. (C) DOC-2-GFP (green, top right panel) localized to the plasma membrane in mature hyphae. Co-localization between DOC-2-GFP with MAK-2-mCherry (red, bottom left panel) was not observed (overlay, bottom right panel). (D) tef-1 driven GFP-DOC-2 (green) showed localization to puncta in germlings, but that do not show oscillation during chemotropic interactions. Scale bars: 10 μm.

https://doi.org/10.1371/journal.pbio.1002431.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. DOC-1-GFP co-oscillates with MAK-2 during chemotropic interactions.

(A) Germlings expressing DOC-1-GFP (green) were paired with germlings expressing MAK-2-mCherry (red). When MAK-2-mCherry accumulated at one CAT tip (arrows left and right panel), DOC-1-GFP was absent from the tip of the interacting germling. When DOC-1-GFP accumulated at the second CAT tip (arrow middle panel), MAK-2-mCherry was absent from the first CAT tip, indicating that DOC-1-GFP showed identical oscillation dynamics to MAK-2-mCherry. (B) Germlings expressing DOC-1-GFP (green) were paired with germlings expressing SOFT-mCherry (red). When SOFT-mCherry accumulated at one CAT tip, DOC-1-GFP accumulated at the CAT tip of its interacting partner (arrows left and right panel). When SOFT-mCherry was absent from the first CAT tip, DOC-1-GFP was absent from the second CAT tip (middle panel), indicating that DOC-1-GFP showed opposite oscillation dynamics to SOFT-mCherry. See S5B Fig for oscillation intervals. (C) Co-localization and co-oscillation of DOC-1-GFP (left panel) with MAK-2-mCherry (middle panel) in heterokaryotic germlings undergoing chemotropic interactions (overlay, right panel). Bottom panel: Graphic representation of DOC-1-GFP and MAK-2-mCherry signals at the tips of conidial anastomosis tubes in germlings undergoing chemotropic interactions. y-axis shows the ratio of relative fluorescence intensity (RFI) in the interacting zone as compared to background. x-axis shows time (min). Note the co-oscillation of both DOC-1-GFP and MAK-2-mCherry in both germlings following fusion, as shown previously for MAK-2 and SOFT [45].

https://doi.org/10.1371/journal.pbio.1002431.g005

In hyphae, DOC-1-GFP localized to puncta that also oscillated at the hyphal tip during chemotropic interactions prior to hyphal fusion (similar to MAK-2 and HAM-5; [45]), as well as to septa (Fig 4B; S4 Movie). To determine whether DOC-1-GFP localization was dependent on MAK-2, we expressed DOC-1-GFP in a strain deleted for mak-2. The Δmak-2 (doc-1-gfp) strain showed a typical Δmak-2 phenotype, including lack of chemotropic interactions and cell fusion. Additionally, DOC-1-GFP showed cytoplasmic and vacuolar localization and never localized to puncta, although localization to septa was retained in this strain. Thus, DOC-1 is a component of the MAK-2 signaling complex and co-oscillates with this complex during chemotropic interactions.

A doc-2-gfp allele regulated by the ccg-1 promoter localized to the plasma membrane and septa in mature colonies (Fig 4C). When a fully functional gfp-doc-2 allele was placed under the regulation of the tef-1 promoter (ccg-1 has low expression levels in germlings compared to tef-1; [47]), GFP-DOC-2 localized to puncta in germlings (Fig 4D). However, oscillation of DOC-2 during chemotropic interactions in either germlings or in fusion hyphae was never observed.

DOC-1/DOC-2 Mediate Communication Group Discrimination by Preventing Reinforcement of Signaling

The oscillation of signaling components is necessary to maintain chemotropic interactions; inhibition of MAK-2 kinase activity obliterated oscillation and chemotropic interactions in communicating germlings [44]. The function of DOC-1/DOC-2 in communication group discrimination could be to prevent initiation of signaling, or to prevent reinforcement of signaling, which is hypothesized to be required for sustained chemotropic interactions [49,61]. To differentiate between these two hypotheses, we analyzed localization of MAK-2-GFP in Δdoc-1 Δdoc-2 germlings when they were in close proximity to CG2 germlings (JW262), where communication frequency is ~50% (Fig 3A). Prior to chemotropic interactions, localization of MAK-2-GFP in Δdoc-1 Δdoc-2 germlings was observed at the tip when in close proximity to CG2 germlings (Fig 6A). However, chemotropic interactions were only established if oscillation of MAK-2-GFP occurred (Fig 6B; S5 Movie). If oscillation of MAK-2 was not maintained, chemotropic interactions between Δdoc-1 Δdoc-2 and CG2 germlings were abolished. These observations suggest that DOC-1/DOC-2 do not function at the recognition stage of germling interactions, but instead function to mediate communication group discrimination at the point where robust oscillation of the MAK-2 signaling complex to the tips of conidial anastomosis tubes is reinforced, and which is essential for further chemotropic interactions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Enforcement of oscillation of signaling proteins is necessary to establish chemotropic interactions regardless of CG.

(A) MAK-2-GFP localized to the tips of conidial anastomosis tubes in a Δdoc-1 Δdoc-2 (CG5) germling when in proximity to a CG2 (JW262) germling (top panel). However, further oscillation of MAK-2-GFP did not occur, and chemotropic interactions between the Δdoc-1 Δdoc-2 (CG5) germling and JW262 (CG2) germling were not established. Bottom panel: Graphic representation of MAK-2-GFP signals at the tip of a Δdoc-1 Δdoc-2 (CG5) germling in proximity to a JW262 (CG2) germling. y-axis shows the ratio of relative fluorescence intensity (RFI) in the interacting zone as compared to background. X-axis shows time (min). (B) MAK-2-GFP in a Δdoc-1 Δdoc-2 germling shows continued oscillation to the CAT tip if chemotropic interactions are established between Δdoc-1 Δdoc-2 (CG5) and JW262 (CG2) germlings (top panel). Bottom panel: Graphic representation of MAK-2-GFP signal at the CAT tip of Δdoc-1 Δdoc-2 (CH5) germlings undergoing chemotropic interactions with germlings of JW262 (CG2). y-axis shows the ratio of RFI in the interacting zone as compared to background. x-axis shows time (min) (See S5 Movie for second example).

https://doi.org/10.1371/journal.pbio.1002431.g006

The necessity of coordinated and out-of-phase oscillation of MAK-2 complex with SOFT for successful communication [44] suggested that the DOC proteins might influence the oscillation interval (8–10 min in CG1 strains) and that differences in oscillation timing might determine communication group affiliation. To test this hypothesis, we compared the oscillation timing of SOFT-GFP during chemotropic interactions in germlings from a CG1 strain (FGSC 2489 [so-gfp]) [44] as compared to the CG5 strain (Δdoc-1 Δdoc-2 [so-gfp]). We quantified fluorescence intensities at the tip of communicating germlings and measured the interval between two fluorescence maxima. However, no significant differences between the oscillation intervals of SOFT could be detected for CG1 versus CG5 germlings, suggesting that alteration of oscillation timing was not the basis of communication group phenotype (S5 Fig).

doc-1, doc-2, and doc-3 Show Evidence of Long-Term Balancing Selection

Our data indicate that doc-1 and doc-2 function as helping greenbeard genes, with multiple alleles mediating assortative kind recognition by changing chemotropic behavior by negatively regulating interactions during germling fusion. The finding of five communication groups mediated by five highly divergent haplotypes suggested a relatively ancient origin of the communication locus controlling germling fusion. To test this hypothesis, we first performed phylogenetic analyses of alleles at NCU07190 through NCU07193 in the 26 sequenced wild isolates, as well as alleles at these same loci from a population sample from the distantly related species Neurospora discreta. For NCU07190 and NCU07193, allelic lines from within species were reciprocally monophyletic (Fig 7A), as predicted by theory [62], given the estimated divergence time between N. crassa and N. discreta (7–10 million years ago [63]) and their effective population size (circa 10⁶ and 10⁴ individuals, respectively [51,64]). However, for the three doc genes, no reciprocal monophyly was observed, and N. crassa alleles from the same CGH-associated clade were closer to N. discreta alleles than to N. crassa alleles from another clade, indicating that the age of allelic lines exceeds the age of speciation events—a phenomenon referred to as transspecies polymorphism (Fig 7B). Inferred genealogical histories of doc genes were in fact concordant with differences in patterns of genomic arrangements among communication group haplotypes: alleles from CGH1 to CGH5 (including CGH1A and CGH1B) were in distinct clades for doc-1, and similar topologies were also inferred at doc-2 (although CGH5 strains lack doc-2) and doc-3 (found only in CGH2 and CGH4 strains; Figs 2 and 7B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Phylogenetic analyses of the predicted proteins encoded by genes associated with genomic rearrangements revealed transspecies polymorphisms at DOC-1 and DOC-2/DOC-3.

Coding sequences from 26 N. crassa wild Louisiana isolates and up to eight N. discreta wild isolates were used to build maximum likelihood phylogenetic trees for (A) NCU07190 (top) and NCU07193 (bottom) and (B) NCU07191 (doc-1) and NCU07192 (doc-2; doc-3), using the default pipeline from Phylogeny.fr [99]. Bootstrap values are given for each node. Black bars indicate substitution rates. CGH1 isolates are shown in orange (subgroups A and B are indicated), CGH2 isolates are shown in light green, CGH3 isolates are shown in blue, CGH4 isolates are shown in purple, and CGH5 isolates are shown in dark green. The isolate JW246 did not produce spores and is therefore shown in black. N. discreta isolates are shown in red. Note nesting of N. discreta isolates within N. crassa lineages for NCU07191 (doc-1) and NCU07192 (doc-2; doc-3), but not for NCU07190 or NCU07193 (shown in panel A); see S1 Tree for Nexus file.

https://doi.org/10.1371/journal.pbio.1002431.g007

Transspecies polymorphism is a signature of long-term balancing selection [65], and evidence for balancing selection was also provided by tests of the standard neutral model using Tajima’s D, which measures skewness of the allele frequency spectrum (S3 Table). Tajima’s D values at NCU17048, NCU07190, NCU07193, or NCU07194 did not deviate from neutral expectations (Tajima’s D < 1; p > 0.1), while values of Tajima’s D were high, positive, and significant for doc-1, doc-2, and doc-3 (Tajima’s D > 2; p < 0.01 for doc-1 and doc-2; p < 0.05 for doc-3). These data indicate that balancing selection is acting to maintain polymorphisms at doc-1, doc-2, and doc-3, but its signature is not detectable on surrounding genes.

The finding of long divergent haplotypes under balancing selection at the doc communication locus suggested that recombination rates might be reduced across the region, thereby preventing the migration of variants between allelic lines [65,66]. To test for recombination within the doc region between isolates of different CGHs, we analyzed concordance among genealogies of all genes within the region [67]. These analyses of genealogical concordance within the doc region revealed congruent branching of sequence groups from different CGHs for doc-1, doc-2, and doc-3 over the entire length of the genes, consistent with a lack of recombination between haplotypes from different communication groups (S6 Fig). In contrast, an analysis of genealogical concordance within the doc region among haplotypes defining the same communication group was consistent with multiple recombination events (S7 Fig), except for CGH1 isolates, in which recombination was not observed within the doc-1/2 region between the CGH1A and CGH1B isolates (S6 and S7 Figs). These data suggest that the recombination rate between haplotypes from different communication groups was reduced, probably because of strong selection against recombinants.

Strains and Growth Conditions

Standard protocols for N. crassa can be found on the Neurospora homepage at the Fungal Genetics Stock Center (FGSC, http://www.fgsc.net/Neurospora/NeurosporaProtocolGuide.htm). Strains were grown on Vogel’s minimal medium (VMM [84], with supplements as required) or on Westergaard’s synthetic cross medium for mating [85]. The wild N. crassa strains used in this study (S1 Table) were isolated from Louisiana, US and are available at the FGSC [52,86,87]. Manipulated strains are listed in S4 Table. FGSC 2489 served as parental strain for gene deletions and as a WT-control for all experiments, unless stated otherwise. Single deletion mutants are available at the FGSC [58,88]. The Δdoc-2 mutant deposited at the Neurospora knockout collection showed a flat phenotype, and its conidia had slow germination rates. In a back cross with FGSC 2489, none of the phenotypes co-segregated with hygromycin resistance, indicating that it was due to a secondary mutation. To create the Δdoc-1 Δdoc-2 mutant, a deletion construct was created using the method of fusion PCR [89]. Briefly, ~1 kb of the 5′ regions of doc-1 and doc-2 was amplified by PCR from genomic DNA (S5 Table), and the hygromycin cassette was amplified from the vector pCSN44 [58]. The three fragments were fused in a fusion PCR reaction to create the deletion construct, which was used to transform the Δmus-51 strain of N. crassa [58]. Hygromycin-resistant transformants were analyzed by PCR, and positive strains were backcrossed to FGSC 2489.

Histidine auxotrophic strains for complementation experiments were obtained by crossing the doc deletion strains with FGSC 6103. The plasmid pMF272 (AY598428) was modified to create gfp-fusions to doc-1 and doc-2, which were targeted to the his-3 locus [90]. A 300 bp fragment of the 3′ region of ccg-1 was cloned 3′ of gfp open reading frame as a termination signal using the EcoRI restriction site. Plasmid derivatives with the tef-1 promoter or native promoters were obtained by swapping out the ccg-1 promoter using the restriction enzymes NotI and XbaI. For CG switch experiments, doc-1 and doc-2, including their native promoter and terminator sequences, were amplified from genomic DNA of the isolate P4471 (S5 Table). Using the Gibson assembly, both fragments were cloned into the EcoRI/NotI digested vector pMF272 [91]. All constructs were transformed into FGSC 6103 with selection for His⁺ prototrophy and then crossed into the doc double or single deletion mutants.

Assays for Germling Communication

Each strain was grown on VMM in slant tubes for 4–6 d or until significant conidiation occurred. Conidia were prepared by filtering 600 μl of conidial suspension through cheesecloth. An aliquot of 180 μl of a conidial suspension from one strain was mixed with 20 μl of FM4-64 solution (16 μM), incubated for 15 min, and subsequently washed with 1 ml of ddH₂O. The conidial titer was adjusted to 3×10⁷ conidia/ml. An aliquot of 45 μl of conidial suspension from both strains was mixed, and 80 μl of this final mixture were spread on VMM agar plates (60 x 15 mm). Plates were incubated for 4.5 h at 25°C or 3.5 h at 30°C. Agar squares of 1 cm² were excised and observed with a Zeiss Axioskop 2 equipped with a Q Imaging Retiga-2000R camera (Surrey) using a 40x/1.30 Plan-Neofluar oil immersion objective and the iVision Mac4.5 software. Different strains were either discriminated by GFP or FM4-64 fluorescence, or FM4-64 fluorescence versus no fluorescence, if two wild isolates were analyzed. Communication frequencies were determined for at least 15 fields, depicting a total of at least 100 interactions with three biological replicates.

Heterokaryon Assays

Conidia of strains bearing different auxotrophic markers (his-3, ad-3B, or pyr-4; S4 Table) were harvested as described above. The conidial titer of one strain was adjusted to 3 x 10⁶ conidia/ml, and the conidial titer of the forced communication partner (bearing a different auxotrophic marker) was adjusted to 3 x 10⁵ conidia/ml. A 150 μl spore suspension of both strains was mixed and spread on modified VMM agar that promotes colonial growth (FGSC, http://www.fgsc.net/Neurospora/NeurosporaProtocolGuide.htm). Due to the complementing auxotrophic markers, only heterokaryotic, prototrophic fusion products were able to grow on VMM. Plates were incubated at 30°C for 4 d, when cell-forming units/plate were documented.

Confocal Microscopy

Cellular localization studies were performed with a Leica SD6000 microscope with a 100×1.4 NA oil-immersion objective equipped with a Yokogawa CSU-X1 spinning disk head, a 488 nm laser for GFP fluorescence, and a 563 nm laser for mCherry fluorescence controlled by the Metamorph software (Molecular Devices, Sunnyvale, CA). Conidia from strains expressing fluorophore-tagged proteins were prepared for microscopy as described above. For time-lapse studies, images were taken at 30 s intervals. The software ImageJ (http://imagej.nih.gov/ij/) was used for image processing.

For co-localization studies, heterokaryons were created by inoculating the center of a plate with a mixture of conidia of a strain expressing DOC-1-GFP and a strain expressing MAK-2-mCherry, or a strain expressing DOC-2-GFP and a strain expressing MAK-2-mCherry, respectively (S4 Table). Conidia bearing both GFP and mCherry fluorescent proteins were prepared and imaged as explained above.

Bulked Segregant Analyses and Genome Resequencing

For DNA isolation, strains were grown on VMM agar plates covered with a disk of sterile cellophane at 30°C for 24 h. DNA was purified using the DNeasy Blood & Tissue kit (Qiagen Inc.). Equal amounts of DNA from 46 segregants (66 ng/segregant) were combined and used for library preparations using the TruSeq DNA LT Kit (Illumina). All paired end libraries were sequenced on a HiSeq2000 sequencing platform using standard Illumina operating procedures (Vincent J. Coates Genomics Sequencing Laboratory, Berkeley) to a read length of 100 nucleotides and a minimum mean depth of genome coverage of 71 for the sequenced libraries after filtering for low-quality reads, using the DepthofCoverage program from GATKv2.3–9 [92]. Low-quality reads were removed from the sequencing data using the Fastx toolkit (http://hannonlab.cshl.edu/fastx_toolkit/index.html). The filtered paired ends were regrouped by a custom perl script and mapped to the N. crassa genome FGSC 2489 v12 with the short read aligner Bowtie2.00 [93]. Read groups were added to sorted BAM files with Picard-tools v1.85 (http:broadinstitute.github.io/picard/) and SNP analysis performed with The Genome Analysis Tool Kit v2.3–9 after indel realignment with the RealignerTargetCreator and IndelRealigner programs from the GATK [92]. SNPs were confirmed by viewing the mapped polymorphisms on the Integrative Genomics Viewer v2.3 [94]. The mapped reads for the two parental strains (FGSC 2489 and JW258) plus the mapped reads for the 46 pooled segregants (FGSC 2489 communicators, CG1 or JW258 communicators, CG2) are available at the Sequence Read Archive (SRA) (http://www.ncbi.nlm.nih.gov/sra) (SRA311058).

Sequence Analysis

The doc-1 and doc-2 sequences of N. crassa and N. discreta wild isolates were obtained by a BLAST search [95] using NCU07191 and NCU07192 from FGSC 2489 as a query against de novo sequence assemblies from 26 wild isolates [55]. For DNA sequence comparisons, the pairwise sequence alignment tool EMBOSS Needle from EMBL-EBI was used [96]. Codon alignments were carried out using Macse [97] and visualized and processed using JalView (http://www.jalview.org/). Modified multiple alignments were trimmed using Trimal [98]. Phylogenetic trees were inferred from trimmed alignments using the default pipeline from Phylogeny.fr (Muscle, Gblocks, Phyml [100 bootstraps])[99] and visualized using FigTree1.4 (http://tree.bio.ed.ac.uk/software/figtree/). To obtain DNA divergence statistics, the trimmed codon alignments of doc-1, doc-2, and doc-3 sequences were sorted based on CGH groups. DnaSP5 was used to compute polymorphism and divergence and to test the standard neutral model using Tajima’s D [100]. Partitioned alignments for each locus were created using RAxML [101].

To detect recombination within CGHs in the region around doc-1/2, the program Rdp [67] was used, applying the “all methods mode” with default setting. Sequences of about 30 kbp surrounding doc-1/2 were extracted from de novo genome sequence assemblies. Alignments were made using the program Mafft for each CGH group [102]; gaps were trimmed using Trimal [98].