“Adapted” mutants arise out of independent “loss of function” strains upon long-term cultivation on stress
Apart from molecular biology techniques and next-generation sequencing approaches, classical vegetative growth assays were used to characterize potential functions of genes associated with osmoregulation to study the HOG pathway in M. oryzae. Therefore, we cultivated the previously generated lof mutants ΔMohog1, ΔMopbs2, ΔMossk2, ΔMossk1, ΔMoypd1, ΔMohik1 and ΔMosln1 (20) on different stress-inducing media to compare them with the wildtype strain. Mutants with an inactivated HOG pathway are sensitive to osmotic stress and resistant to the fungicide fludioxonil (20). We noticed in the course of these assays that individual mycelium parts grew out of the sensitive lof mutants ΔMohog1, ΔMopbs2, ΔMossk2, ΔMossk1 and ΔMoypd1 after cultivation for at least four to six weeks under continuous salt stress (Fig. 2).
We isolated these individual mycelium parts in order to separate them as pure cultures ready for further investigations and named them ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted). We have not been able to get adapted strains from the mutants ΔMosln1 and ΔMohik1 in the conditions tested so far. That underpins the hypothesis from our previous studies that the two-component hybrid histidine kinase (HK) MoHik1p and the HK MoSln1p can partially take over the function from each other (23) and, thus, the selection pressure maybe not sufficient for an adaptation event in ΔMosln1 and ΔMohik1. Furthermore, it was not possible for us to isolate adapted strains from HOG pathway-independent osmosensitive Magnaporthe-mutant strains, i.e. ΔMostu1 (transcription factor in cAMP/PKA signaling pathway, ΔMogpd1 (glycerol-3-phosphate dehydrogenase) or ΔMoskn7 (response regulator protein). Of course, we checked in the adapted strains whether the genes originally inactivated in the “parent” lof mutants were still inactivated in order to avoid any possibility of contaminations or confusions about mixed cultures. We did this for all adapted strains using ITS-sequencing and southern blot analyses (supplementary Fig. S1).
In empirical investigations, a combination of different vegetative growth assays was used to characterize the “rapid adaptation frequency” in the different lof mutants. Twenty mutant strains of each of ΔMohog1, ΔMopbs2, ΔMossk2, ΔMossk1, and ΔMoypd1 were grown continuously on solid complete medium (CM) including 1 M KCl or 1.5 M sorbitol as stress-inducing agents. Several adapted strains arose out of the lof mutants in each plate, being able to grow much faster under the stress condition. The adapted strains were then transferred onto CM without stressors for the following two weeks. Subsequently, we transferred the colonies onto repeated stress medium to investigate in which strains the “adaptation” is stable and identified the stably adapted strains (Fig. S2). Notably, we were able to identify more adapted strains from ΔMopbs2 and ΔMoypd1 under salt stress than under sorbitol stress. By contrast, we obtained more adapted strains under sorbitol stress than under salt stress from ΔMossk2 and ΔMossk1 (Supplementary Fig. 2).
Osmoregulation is permanently restored in the adapted mutants
The strains ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted) displayed significant differences in growth speed on salt stress compared to their “parent-strains,” the lof mutants ΔMohog1, ΔMopbs2, ΔMossk2, ΔMossk1 and ΔMoypd1 (Fig. 3 [1]). All lof mutants were strongly sensitive towards 1 M KCl stress, whereas all the adapted strains were less sensitive, similar to the wildtype strain (Fig. 3 [1], lower row, colonies [A]). Particular attention has to be paid to the finding that adapted strains, which were pre-cultivated on normal CM (unstressed conditions) and then transferred back to the repeated salt stress, were found to grow as fast as those taken directly from stress conditions and, thus, being able to adapt to KCl stress immediately (Fig. 3 [1], colonies [C]). In conclusion, the mutations/alterations within ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted) appear to be stable. Similar results to the growth assays on solid media could be observed in liquid cultures upon KCl stress (Fig. 3 [2]) and sorbitol stress (supplementary Fig. S3).
Glycerol is the major compatible solute produced by the adapted strains after salt shock
Intracellular production of compatible solutes was determined by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) and compared to compatible solute production of the lof mutants and the wildtype strain to further investigate the phenomenon of restored osmoregulation in ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted). Hyperosmotic shock was imposed by 0.5 M KCl stress towards the fungal strains, and the intracellular levels of the major osmolytes mannitol, trehalose, arabitol and glycerol were determined. No increases in the mannitol and trehalose levels were detected in the lof mutants and the adapted strains after osmotic shock (data not shown). A slight increase in the mannitol level was detected and no significant change for the trehalose level of the wildtype strain (data not shown). Similar to data from (19), arabitol was found to be the major intracellular compatible solute produced by the wildtype strain after osmotic shock (Fig. 4).
By contrast, the lof mutants ΔMohog1, ΔMopbs2, ΔMossk2, ΔMossk1 and ΔMoypd1 were not able to produce either arabitol or glycerol in significant amounts. Interestingly, it was found that all the adapted strains ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted) responded to hyperosmotic stress by accumulating high amounts of glycerol rather than arabitol (Fig. 4). Based on these observations, we conclude that glycerol may somehow compensate for the lack of arabitol upon salt stress in the adapted strains.
Fludioxonil susceptibility is restored in the adapted mutants
Vegetative growth assays were conducted using minimal medium (MM) and MM including 10 µg/ml fludioxonil to further investigate whether fludioxonil-susceptibility and not only osmoregulation is restored in ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted) (Fig.5).
Apart from the osmoregulation capacity, fludioxonil sensitivity was reconstituted in the adapted strains. The HOG pathway lof mutants were resistant towards the fungicide (Fig. 5, colonies [A], lower row), whereas all adapted strains were susceptible, but not quite as strongly as the wildtype strain (Fig. 5, colonies [B], lower row). Similar to the findings concerning salt stress presented previously in Fig. 3, the adapted strains, which were pre-cultivated in unstressed conditions for a long time, showed the same phenotype as the adapted strains taken directly from stress medium (Fig. 5, colonies [C], lower row).
Fludioxonil sensitivity in the adapted strains is not dependent on the production of arabitol or glycerol
Fludioxonil induces a hyperactivation of the HOG pathway and, thus, a continuous stress response towards high osmolarity. We found, by using HPAEC-PAD analytics, that fludioxonil treatment resulted in the production of arabitol and glycerol in the wildtype strain (Fig. 6). We conducted the experiments with the lof mutants and the adapted strains ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted) to check whether fludioxonil-dependent production of the compatible solutes was altered in the adapted strain. The fungal cultures were grown in liquid CM (2 % glucose) and stressed with 10 µg/ml fludioxonil. As expected, we did not find increased compatible solute production upon fludioxonil-treatment in the lof mutants. Exemplarily, we present the data for ΔMohog1 (Fig. 6).
Interestingly, no increase of arabitol or glycerol production was detectable in the fludioxonil-susceptible adapted strains in the presence of the fungicide (Fig. 6). It has to be pointed out that the metabolic response of the adapted strains after fludioxonil treatment is different compared to the compatible solute production we observed upon KCl treatment (Fig. 4). All the adapted strains ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted), ΔMossk1(adapted) and ΔMoypd1(adapted) responded to salt stress by accumulating high amounts of glycerol, whereas this is not the case under fludioxonil stress. In conclusion, fludioxonil sensitivity in the adapted strains does not appear to be dependent on compatible solute production.
Reestablished osmoregulation does not complement reduced virulence of the lof mutants
The lof mutants of the HOG pathway in M. oryzae were found to be reduced in virulence towards rice plants compared to the wildtype strain. Interestingly, ΔMohog1(adapted), ΔMopbs2(adapted), ΔMossk2(adapted) and ΔMossk1(adapted) were not found to be as virulent as the wildtype strain, and rather less virulent than the lof mutants (Fig. S4). We were not able to conduct the pathogenicity assays regarding ΔMoypd1(adapted), since the mutant failed to produce conidia exactly like ΔMoypd1 (20).
There are no relevant structural variations on DNA-level in the genomes of adapted strains
We had a deeper look at the genomes of ΔMohog1 compared to ΔMohog1(adapted) in order to find out the cause of that rapid adaptation in the adapted strains. We also added data from the genome sequencing of ΔMopbs2(adapted) to strengthen the analysis and narrow down the outcome of putative candidate genes showing structural variations in the adapted strains. Single nucleotide variations (SNVs) and short indels were detected for ΔMohog1(adapted), ΔMopbs2(adapted) and ΔMohog1 in comparison with the reference sequence of the wildtype strain. The resulting variants were further annotated based on their chromosomal location and biological effects, such as synonymous/non-synonymous single-nucleotide polymorphisms (SNPs), upstream/downstream, untranslated regions (UTRs) and intergenic regions (Additional file 1 - SVs summary). The ratio of transition and transversion was also calculated for single nucleotide variation. Over 80 % of all SNPs detected were found to be located outside exons and a significant enrichment in regions adjacent to exons and UTRs was detected. Furthermore, in silico protein modelling suggested that several non-synonymous SNPs are probably direct targets of selection, as they lead to amino acid replacements in functionally important sites of proteins. Hence, the structural variation discovery analysis of small-scale (< 20 bp) and large-scale variations (> 20 bp) such as frameshift, stop codon insertion resulted in a list of three genes (five putative gene variants) in the overlap of ΔMohog1(adapted) and ΔMopbs2(adapted), but none of them leads to a protein effect. The variations are only transitions not changing the amino acid composition of the corresponding proteins. Furthermore, we checked the homologous gene loci of the known yeast suppressor mutation genes RSG1 (RHB1) (24), SOO1 (25), SGD1 (26) and PMK1 (KSS1) (27) intensively. Within all these loci, we could not identify structural variations in the genome of the adapted strains ΔMohog1(adapted) and ΔMopbs2(adapted).
Since regulatory gene elements, such as promoters, are of prior interest regarding their direct influence on gene expression alteration, we decided to investigate the presence of genetic structural variations in the putative promoter regions of all annotated genes in the M. oryzae genome. The promoter region was defined as the region on the genomic DNA 1500 bp upstream of each annotated gene start codon. We performed a structural variation discovery analysis of small-scale (< 20 bp) and large-scale variations (> 20 bp) resulting again in no significant variations in the overlap of the promoter regions of ΔMohog1(adapted) and ΔMopbs2(adapted) (Fig. S5 [A]). We considered here only insertions and deletions as a “polymorphism type” and insertions, truncations and frame-shifts as a “protein effect.”
Differential transcriptomic profiles in the adapted strains in response to osmotic stress
Next-generation sequencing analysis of RNA samples from the wildtype strain, ΔMohog1 and ΔMohog1(adapted) before and after 25 min salt stress (0.5 M KCl) should present insights into transcriptional changes which may be responsible for the adaptation phenomenon observed (methods for cultivation before RNA-isolation, see (28)). A principal component analysis was performed to characterize the relationship between the strains analyzed. The processed transcriptome data of ΔMohog1 (untreated), ΔMohog1 (25 min 0.5 M KCl-stress), ΔMohog1(adapted) (untreated) and ΔMohog1(adapted) (25 min 0.5 M KCl-stress) appears to form distinct clusters within each sample group of strain samples investigated (Fig. 7A).
Since the principal component analysis revealed distinct differences on transcriptomes of corresponding strains, we decided to follow up by clustering the genes according to their expression followed by construction of a co-expression network. In the clustering analysis, represented by a heat map visualization, the fact that the transcript value is strongly different in ΔMohog1(adapted) compared to the WT 70-15 and ΔMohog1, even in untreated conditions, is clearly visible (Fig. 7 [B]). Cluster 6 is significantly exclusively upregulated and cluster 1 is harboring exclusively down-regulated genes in ΔMohog1(adapted), whereas exclusively up-regulated genes in the WT 70-15 could be found in cluster 2 (Fig. 7 [B]).
Glycerol metabolism-associated genes are affected in the adapted strains
To investigate, whether the groups of differentially expressed genes (DEGs) are functionally related, we performed a gene ontology (GO) enrichment analysis to determine the biological functions associated to them. Generally, this approach helps to highlight groups of genes with coherent biological functions that are presumably acting in coordination in response to salt stress. The clustering analysis was employed over the whole gene co-expression network and in a selected subset or cluster of interest. The result indicated that there were several significantly enriched terms of DEGs, from which the most representative are “lipid biosynthetic process,” “glutamine family amino acid metabolic process,” “carbohydrate transport,” “dephosphorylation” and “carboxylic acid biosynthetic process” (Fig. 8).
This enrichment analysis suggests that the highly connected genes from the largest cluster in the network are involved in the regulation of complementary processes triggered by salt stress.
In order to follow up the results of glycerol-production in the adapted strains, we further investigated most of the genes potentially contributing to the production, metabolism or transport of glycerol. However, we checked these genes presented resulting in a list of homologous genes potentially related to the production, metabolism or transport of glycerol from a database- and literature-based approach (Tab. 1)
Tab. 1: Glycerol biosynthesis and osmotic stress response-related genes being differentially expressed across the strains analyzed. The table shows the glycerol biosynthesis and osmotic stress response-related genes retrieved being differentially expressed across the strains analyzed. The colors used in the table indicate the percentiles projected on the entire amounts of the transcripts counted (green stands for 90 %, white for 50 % and red for 10 %).
The analysis resulted in a set of candidate genes which were found to be upregulated in both the salt stress samples of the ΔMohog1(adapted) and the wildtype strain, whereas these genes were not regulated in the lof mutant ΔMohog1 (Tab. 1, yellow marked). Among these candidates, we identified genes encoding the glycerol H+-symporter MoSlt1p (MGG_09852), one phosphoglycerate mutase (MGG_06642,), one glycerol-3-phosphate dehydrogenase (MGG_00067 (MoGpd1p)) and one phosphatidyl synthase (MGG_00099 (MoHad1p)).
The HOG pathway is not responsible for adaptation in ΔMohog1(adapted)
We searched for possible interactions between MoHog1p and other osmotic stress responsive or associative gene products using the SMART website to find links between putative interaction partners of the MAPK MoHog1p in the adapted strains. The genes identified were finally used to inspect their expression patterns within our set of DE genes (STRING analysis, Fig. 9 [A]).
As expected, the transcript level of the responsive genes belonging to HOG pathway were found to be highly upregulated in case of the wildtype strain in response to osmotic stress (Fig. 9 [B]). Thus, among the genes with the most abundant transcripts were MGG_01822, MGG_08212 and MGG_08547 encoding HOG1, a BZIP transcription factor and CAMK1 kinase, respectively (Fig. 9 [B], [C]). Meanwhile, none of the HOG1-associated genes, except for MGG_06759 encoding a heat shock protein, were transcriptionally active in ΔMohog1 and in ΔMohog1(adapted) (Fig. 9 [B]). These results have been validated by means of qPCR (Fig. S6, supplementary). That leads to the conclusion that other mechanisms operating independently/outside of the HOG pathway may be responsible for the phenotype observed in the adapted strains.