Fungal fruiting body formation is a complex process that requires coordinated patterns of gene expression in time and space. Even though a number of genes that are essential for this process have been isolated from several model organisms, no unifying theory yet explains the spatio-temporal succession of developmental events leading to the mature fruiting body
. One way to learn more about the genes that are active during this process is to look at genome-wide expression patterns at different developmental stages, but this is difficult in many ascomycetes, because fruiting bodies are often rather small (< 500 μm for the mature fruiting body) and difficult to separate from surrounding, non-reproductive hyphae. In the present study, we used LM and RNA-seq to analyze gene expression in protoperithecia from the model organism S. macrospora. To the best of our knowledge, this study is the first time that a combination of these methods has been used for the analysis of fungal gene expression. Our study demonstrates that young fruiting bodies of S. macrospora can be isolated using LM, and RNA extracted from these samples in sufficient amounts for RNA-seq after two rounds of linear amplification. The amplification process largely conserves expression ratios, as demonstrated by comparing the expression of selected genes prior to amplification with the RNA-seq results, which is consistent with results from other organisms where linear RNA amplification was used to prepare samples for microarray hybridization
[19, 20]. We also found some overlap with prior microarray experiments in which we compared gene expression in vegetative and sexual mycelia; however, in these experiments we used mycelia grown in defined medium
, whereas for RNA-seq analysis, RNA from mycelia grown in defined medium and cornmeal medium were pooled. Therefore, some differences between these experiments might be due to different growth conditions ( Additional file
1 Table S2). Furthermore, we demonstrated that the regulatory regions of ppg1 can drive expression of an egfp reporter gene in protoperithecia, as predicted by the RNA-seq analysis. In addition, fluorescence microscopy analysis revealed distinct expression patterns for ppg1 in the outer layers of the protoperithecium. This finding might be consistent with a hypothesis that has been put forward for N. crassa that predicts that pheromones are not only signaling molecules that enable the recognition of mating partners, but that they also play a role in the attachment (“conglutination”) of hyphae forming the rigid outer perithecial wall
. A role for pheromones as “molecular glue” might explain the expression of ppg1 in cells of the protoperithecial outer layers in S. macrospora.
In a study of gene expression in several tissues of different metazoans, Hebenstreit et al. found that genes can be grouped into two classes, namely genes with high and low expression, independent of tissue type, species or type of experiment (microarray analysis or RNA-seq)
. This classification resulted in two distinct peaks when plotting the distribution of gene expression levels. We wondered whether this distribution might also be found in fungi, but plots of the distribution of gene expression levels showed different patterns for our data ( Additional file
1 Figure S10, Additional file
4). We observed a single main peak in both vegetative and sexual mycelium, whereas the frequency distribution in wild-type and pro1 protoperithecia could be dissected into three peaks. This difference indicates that, in contrast to metazoans, fungal genes might not generally fall into two main classes of expression. One reason might be that in the case of sexual and vegetative mycelium, pooled RNA samples were used from mycelia grown in different types of media. These mycelia might express different sets of genes at high and low levels, and such a mixture would drive overall expression frequencies towards intermediate values
, resulting in a single peak as observed. However, the multiple peaks for protoperithecia cannot be explained by a mixture of different samples. The analysis by Hebenstreit indicated that the genes from the high expression group constitute the active and functional transcriptome of the cell, whereas the genes from the low expression group show “leaky” expression
. Our data indicate that the situation in fungi might be different, but further analysis will be needed to clarify this point.
In previous studies, we used cross-species microarray hybridization to hybridize S. macrospora targets on N. crassa cDNA or oligonucleotide microarrays
[8–11]; however, in these analyses, less than 50% of all genes on the arrays gave a significant signal. The use of RNA-seq dramatically improves detection levels, with more than 90% of all genes being detected in at least one of the sequenced samples. Also, the comparison of gene expression revealed that the overall expression in sexual mycelium is more similar to that of vegetative mycelium than protoperithecia. This finding indicates that gene expression in the sexual mycelium is most likely driven to a large extent by genes expressed in the non-reproductive hyphae making up the bulk of the mycelium and that, in order to study genes specifically expressed in developing fruiting bodies, the microdissection method applied here provides a much better spatial resolution and a much more detailed and specific picture of gene expression during development. Especially weakly expressed, fruiting body-specific genes would most likely not be detected as differentially expressed (or at all) in an expression study using only sexual mycelium. Previous approaches for isolating fruiting bodies for gene expression studies were performed in N. crassa and F. graminearum, using EST sequencing with RNA from mature fruiting bodies
, or by analyzing different stages of fruiting bodies by microarray hybridization
[6, 7, 13]. The analysis by Hallen et al.
 was performed with Affymetrix GeneChips for F. graminearum, and signals were detected for nearly 80% of all transcripts, whereas the EST analysis was limited by a comparatively low sequencing depth, and in the other two microarray studies
[6, 13], only 10% of all genes gave signals or were represented on the arrays. In all studies, fruiting bodies were harvested by scraping developing structures from a plate, and these preparations might contain an undetermined amount of non-fruiting body mycelia, especially in the early stages of development when fruiting body precursors are small. Therefore, the present analysis of microdissected protoperithecia allowed the analysis of gene expression solely in these structures for the first time. An additional advantage of RNA-seq is that the data can be used also for annotation purposes and, in the case of S. macrospora, allowed the modeling of more than 50% of the UTRs, and the improvement of exon-intron structures for about 1,000 genes (~ 10% of the predicted genes in the genome).
The analysis of gene expression ratios and the 500 genes with the highest number of reads in each of the four sequenced samples showed that expression in protoperithecia from the wild-type and mutant pro1 is more similar to each other than to either vegetative or sexual mycelium, indicating that the transcriptional landscape of protoperithecia is distinct from that of non-reproductive mycelium. However, there are also significant differences between protoperithecia from the wild-type and the sterile mutant pro1 that can form protoperithecia, but not mature fruiting bodies. More than 400 genes were significantly up- or downregulated in pro1 protoperithecia compared to wild-type protoperithecia, and therefore might be direct or indirect targets of PRO1. Among the genes that are dependent on pro1 for correct expression in protoperithecia are the pheromone precursor genes, several genes that might be involved in perithecial wall morphogenesis, and a number of transcription factors. Previous analyses identified several mutants in which the pheromone precursor genes are differentially regulated in sexual mycelium compared to the wild-type
[8, 10, 11, 60]; however, no information was available about the spatial regulation of the expression of developmental genes prior to this study. One might hypothesize that pro1 is involved in balancing the expression of genes involved in the formation of the rigid perithecial wall because the pheromone precursor genes and several other genes predicted to be involved in cell-wall biosynthesis are up-regulated in pro1 protoperithecia (Table