Strain Degeneration in Pleurotus ostreatus: A Genotype Dependent Oxidative Stress Process Which Triggers Oxidative Stress, Cellular Detoxifying and Cell Wall Reshaping Genes

Strain degeneration has been defined as a decrease or loss in the yield of important commercial traits resulting from subsequent culture, which ultimately leads to Reactive Oxygen Species (ROS) production. Pleurotus ostreatus is a lignin-producing nematophagous edible mushroom. Mycelia for mushroom production are usually maintained in subsequent culture in solid media and frequently show symptoms of strain degeneration. The dikaryotic strain P. ostreatus (DkN001) has been used in our lab as a model organism for different purposes. Hence, different tools have been developed to uncover genetic and molecular aspects of this fungus. In this work, strain degeneration was studied in a full-sib monokaryotic progeny of the DkN001 strain with fast (F) and slow (S) growth rates by using different experimental approaches (light microscopy, malondialdehyde levels, whole-genome transcriptome analysis, and chitosan effect on monokaryotic mycelia). The results obtained showed that: (i) strain degeneration in P. ostreatus is linked to oxidative stress, (ii) the oxidative stress response in monokaryons is genotype dependent, (iii) stress and detoxifying genes are highly expressed in S monokaryons with symptoms of strain degeneration, (iv) chitosan addition to F and S monokaryons uncovered the constitutive expression of both oxidative stress and cellular detoxifying genes in S monokaryon strains which suggest their adaptation to oxidative stress, and (v) the overexpression of the cell wall genes, Uap1 and Cda1, in S monokaryons with strain degeneration phenotype indicates cell wall reshaping and the activation of High Osmolarity Glycerol (HOG) and Cell Wall Integrity (CWI) pathways. These results could constitute a hallmark for mushroom producers to distinguish strain degeneration in commercial mushrooms.


Introduction
Filamentous fungi frequently degenerate during their maintenance in culture by showing loss of the ability to reproduce sexually or asexually [1]. Degeneration of fungal cultures has been related to a cellular accumulation of Reactive Oxygen Species (ROS). In Aspergillus nidulans mycelia with fluffy sectors, cytochrome c release, calcium overload, and upregulation of apoptotic genes revealed a link between oxidative stress and fungal culture degeneration [2]. Moreover, in Metarhizium anisopliae, sectorial cultures showed syndromes associated with aging, upregulation of genes involved in deoxidation and selfprotection (i.e., Heat-Shock Proteins, HSPs), physiological adaptations such as cell structure reorganization, and the activation of different signaling pathways. [3]. In Volvariella volvacea, an edible basidiomycete, prolonged subculturing of mycelia for six generations led to a the progeny into two groups: fast-(F) and slow-growing (S) monokaryons [52]. Genome sequencing of both protoclones (Joint Genome Institute, JGI) revealed a high degree of synteny between them. A whole-genome transcriptome analysis profile revealed 11,820 genes and a substantial population of transposable elements. The sequence of both protoclones also made possible the identification of Single Nucleotide Polymorphisms SNPs [55], which permitted identifying the parental origin allele in monokaryons derived by meiosis from dkN001. This information enabled us to pose questions and design experiments to uncover the underlying genomic differences between fast-and slow-growing monokaryon strains, which showed strain degeneration.
For that purpose, in this work, we analyzed two subpopulations of eighteen fullsib monokaryons obtained from the dikaryotic strain DkN001 in 2016, as well as their monokaryotic protoclones MkPC9 and MkPC15. The progeny was clustered according to their growth rates as F-and S-subpopulations. Growth rates, biomass production, and light microscope studies were performed. A correlation between monokaryon genotype versus phenotype was carried out in both subpopulations to determine whether any relationship existed to warrant their performance. Finally, whole-genome transcriptome analyses of the full-sib monokaryons and their parental protoclones were carried out. The constitutive expression of two genes, Uap1 (a UDP-N-acetylglucosamine diphosphorylase, EC 2.7.7.83) and Cda1 (a chitin deacetylase, EC 3.5.1.41), both involved in cell wall reshaping in slowgrowing monokaryons, moved us to study the effect of chitosan, a biocidal compound with important medicine and agriculture properties [56], which drastically reduced the growth rate in these monokaryons.

Fungal Strains and Growth Conditions
The P. ostreatus dikaryotic strain DkN001 (Spanish Type Culture Collection accession CECT20600), and the monokaryotic protoclones MkPC9 (CECT20311, fast-growing) and MkPC15 (CECT20312, slow-growing) obtained by dedikaryotization of the strain DkN001 were used in this work [54]. The mating of MkPC9 and MkPC15 restores the dikaryotic strain DkN001. A progeny of 60 monokaryons derived from single spores obtained after meiosis of the strain DkN001, was isolated in 2016. All monokaryons were cultured on Malt Extract Solid Medium (MESM: malt extract, 20 g/L; and bacteriologic agar, 15 g/L). Linear growth measurements were performed until mycelia thoroughly colonized the 9 mm diameter Petri dishes. The growth rate value of each strain is the average of three replicates. Samples for biomass production, RNA-seq, and RT-qPCR experiments were obtained from the protoclones, MkPC9 and MkPC15, and monokaryons of the Fast (F) (Mk01, Mk02, Mk06, Mk08, Mk13, Mk20, Mk24, Mk27, Mk28) and Slow (S) (Mk61, Mk62, Mk69, Mk74, Mk75, Mk76, Mk77, Mk82, Mk83) subpopulations. Three replicates of each strain were grown for six days in Erlenmeyer flasks containing 100 mL of SMY medium (1% Sacarose, 1% Malt Extract, and 0.4% Yeast Extract), and incubated in the dark at 24 • C under orbital shaking (125 rpm) to allow growth adaptation to the culture medium. The adapted cultures were homogenized, and a 15 mL inoculum was used to inoculate flasks containing 135 mL of liquid SMY medium and this was incubated in static Submerged Fermentation (SmF) in darkness for seven days at 24 • C. Mycelia were collected, weighed, frozen in liquid nitrogen, and stored at −80 • C for later use. The correlation between the linear growth rate of the monokaryons on Petri dishes and biomass production was evaluated using the Spearman correlation statistic ρ. It was assumed that both variables were correlated when ρ was >0.65 and the p-value was <0.05.

RNA Extraction, Library Construction, and Sequencing
For RNA analysis, three biological replicates of each strain were grown in SMY as described above. Each replicate was split into two halves: one of them was stored at −80 • C for RT-qPCR experiments, and the second half of replicates were pooled, ground in a sterile mortar using liquid nitrogen, and stored at −80 • C to be used for transcriptome analysis.
A fungal RNA E.Z.N.A. kit (Omega Bio-Tek, Norcross, GA, USA) was used to obtain the total RNA, the quality of which was determined by electrophoresis on 1% (w/v) agarose gels. Technical RNA duplicates were performed to assess RNA concentration measured by Qubit ® 2.0 fluorometer using a Qubit ® RNA Assay kit (Invitrogen, Life Technologies Corporation, USA) and Bioanalyzer (version 2100). All samples had 260/280 ratios greater than 2.0 and RNA integrity numbers (RIN) ≥ 8.8. All mRNA libraries were constructed using the TruSeq ® RNA sample (Illumina ® , Inc., San Diego, CA, USA), following the manufacturer's instructions. The sequencing was performed with an Illumina HiSeq 2000 system using 75 bp paired-end reads.

SNPs and Crossover Detection
Single Nucleotide Polymorphisms (SNPs) between protoclones MkPC9 and MkPC15 were detected using RNA-seq data. Alleles of MkPC9 and MkPC15 were identified by reciprocal BLAST (Basic Local Alignment Search Tool, [59]). Each generated alignment was analyzed to identify genomic polymorphisms between the two protoclones and saved in a VCF file. RNA-seq data of the 18 monokaryons of this study were mapped to the reference genome (MkPC15) using the STAR program [58]. Each mapping was done with the VCF file using the GATK HaplotypeCaller program [60,61] to identify each SNP and determine the allele origin.

Correlation Analysis between Growth Rate and Gene Expression (RPKMs)
The correlation between gene expression (RPKMs) and growth rate (mm/day) was studied using the Spearman correlation statistic, ρ. Values of ρ greater than 0.65 indicate a positive correlation (PC) between a specific gene and growth rate, whereas values lower than −0.65 indicate a negative correlation (NC). The False Discovery Rate (FDR method, [62]) cut off used was <0.05 and the p-value was <0.005. To identify the genes differentially expressed (DEGs) between F-and S-subpopulations, the abundance of the transcripts was calculated according to the Reads per Kilobase of exon per Million mapped reads (RPKMs) method. R statistical package software EdgeR (Empirical Analysis of Digital Gene Expression in R, http://www.bioconductor.org/packages/2.12/bioc/html/edgeR. html, accessed on 11 October 2018) was used for differential expression analysis [63]. Only genes with a Fold Change (FC) of >1.5, a FDR cut off of <0.05 and p-value of <0.005 were used for the analyses.

Functional Analysis of Genes
Correlated and DEG genes were functionally annotated using OmicsBox software [64], and their Eukaryotic Orthologous Group (KOG) classification was performed using EggNOGmapper (OmicsBox-Biobam Bioinformatics). Genes were blasted against the NCBI nonredundant protein database, and the InterProScan was used to search domains. Allocation by GO domains into the different categories: Biological Process (BP), Cellular compo-nent (CC), and Molecular Function (MF), are shown according to generic terms in the OmicsBox software.

RT-qPCR Analysis
The relative expression level of selected genes was analyzed to validate RNA-seq data via quantitative reverse transcription PCR (RT-qPCR). Total RNA (800 ng per sample) was reverse transcribed into cDNA using an iScript cDNA synthesis kit (Bio-Rad, Laboratories, Inc. Hercules, CA, USA). Reverse transcription (RT) was carried out in a thermal cycler (MJ Research, Inc. Waltham, MA, USA) using the following program: 5 min at 25 • C, 30 min at 42 • C, and 5 min at 85 • C. RT-qPCRs were performed in a CFX96 real-time system (Bio-Rad, Laboratories, Inc.) using SYBR green dye to detect product amplification. Each reaction mixture contained 10 µL iQ SYBR green supermix (Bio-Rad, Laboratories, Inc.), 2 µL of 3 µM forward and reverse primers, 1 µL of a 1:20 dilution of the RT product, and 5 µL of sterile water. The amplification program consisted of 5 min at 95 • C and 40 cycles of 15 s at 95 • C, 30 s at 63 • C, and 15 s at 72 • C, followed by a final melting curve analysis in which the temperature was increased in increments of 0.5 • C every 5 s in a linear gradient from 65 to 95 • C. Three biological replicates were used for each monokaryon. Relative gene expression was determined by the 2 −∆∆Ct method using the GenEx software for processing and analysis of qPCR data [65]. All Primer pairs used in this work are shown in Table S1, and their amplification efficiencies were greater than 95%. qPCR data were normalized using three reference genes (Table S1) according to Castanera et al. [66].

ROS Detection Assay
Monokaryons were grown on MESM Petri dishes until they reached a diameter of 30 ± 3 mm. For superoxide radical (O2 − ) detection, the colonies were flooded with 10 mL of staining solution containing 5 mM 3-(N-morpholino) propane sulfonate NaOH buffer, pH 7.6, and 2.5 mM of Nitro Blue Tetrazolium (NBT, Sigma-Aldrich, St. Louis, MO, USA) and incubated at 24 • C for 30 min. The staining solution was discarded, and plates were incubated for 1.5 h at 24 • C in the dark.
For H2O2 detection, a similar procedure to the NBT was applied using a staining solution of 100 mM potassium phosphate buffer, pH 6.9, 2.5 mM Diaminobenzidine tetrachloride (DAB, Sigma-Aldrich), and 5 purpurogallin units/mL of horseradish peroxidase (Type VI, Sigma-Aldrich) freshly prepared and shielded from light to avoid spontaneous photo-oxidation. The reaction produced a brownish polymer in the mycelium.

Lipid Peroxidation Assay
Lipid peroxidation was measured in all the monokaryons of the F-and S-subpopulations by the thiobarbituric acid (TBA) test, according to Hodges et al. [67]. Mycelia were handled as previously described for RNA-seq experiments, frozen and ground in a sterile mortar in the presence of liquid nitrogen, and homogenized in 1 mL prechilled 0.1% (w/v) trichloroacetic acid (TCA) solution. Homogenates were centrifuged at 20,000× g for 5 min at 4 • C, and 750 µL of the supernatant were mixed with 750 µL of the Reagent Solution II (RSII: RSI (20% w/v TCA and 0.01% v/v Butylated hydroxytoluene (BHT)) + 0.65% w/v TBA), mixed vigorously and incubated at 95 • C for 25 min. The reaction was stopped by placing the tubes in ice. Samples were centrifuged as above, and the supernatants were transferred to new tubes. Absorbances were read with a Shimadzu spectrophotometer Uvi-1900 (Shimadzu Corporation, Kyoto, Japan) at 440 nm (sugar absorbance), 532 nm (maximum absorbance of pinkish-red chromagen, the product of the reaction of malondialdehyde (MDA) with TBA) and 600 nm (turbidity). The reference solution consisted of 0.1% (w/v) TCA. MDA equivalents were estimated as follows: MDA equivalents (nmol/mL) = [(A − B)/157000] × 106 where A = [(Abs 532RSII − Abs 600RSII)] and B = [(Abs 440RSII − Abs 600RSII) × 0.0571] MDA equivalents (nmol/g Fresh Weight) = MDA equivalents (nmol/mL) × total volume of the extracts (mL/g) FW.

Chitosan Treatments
Sensitivity to chitosan was analyzed in monokaryons of F-and S-subpopulations grown on Petri dishes containing MESM amended with 0.5, 1, and 2 mg/mL chitosan concentrations. Medium molecular weight chitosan (70 kDa) with an 82.5% deacetylation degree (T8s; Marine BioProducts GmbH; Bremerhaven, Germany) was used. Chitosan was prepared according to Palma-Guerrero et al. [34]. The chitosan effect was evaluated by measuring the reduction in the mycelium growth rate of each strain grown for 10 days in Petri dishes containing different chitosan concentrations. Three replicates of each monokaryon were evaluated. Data obtained were analyzed with IBM SPSS statistics 27.0 software (IBM Corp. Released 2020. IBM SPSS Statistics, Version 27.0. Armonk, New York: IBM Corp) using the Scheffe post hoc statistic to compare all possible simple and complex pairs of means between the different doses and control of each monokaryon. Mycelia were harvested to be used in RT-qPCR experiments as described above.

Protein Purification and Immunoblot Analysis
Phosphorylation of Mitogen-activated Protein Kinases (MAPKs) was analyzed in monokaryons of F-and S-subpopulations. Mycelia were obtained as described previously for RNA-seq experiments. Briefly, frozen mycelia maintained at −80 • C were ground in a mortar, and resuspended in protein extraction buffer containing 10 mM HEPES, 50 mM KCl, 1 mM EGTA, 1 mM MgCl 2 , Protease Inhibitor Cocktail tablets (Roche Diagnostics, SL, Basel, Switzerland) and PhosSTOP Phosphatase Inhibitor Cocktail tablets (Roche Diagnostics, SL). Samples were centrifuged to pellet debris, and the protein concentration of the soluble fraction was measured by Bradford assay (Bio-Rad Laboratories, Inc.). Twenty-five micrograms of total protein were separated in a 10% SDS-PAGE gel and transferred to nitrocellulose membranes for Western blot analysis using a Trans-Blot Turbo Mini Nitrocellulose Transfer Pack following the manufacturer's instructions (Bio-Rad Laboratories, Inc.). Phosphorylation of TEY and TGY motifs of MAPKs were detected with phosphor-p44/42 (ERK1/2 homologous) and -p38 (Hog1 homologous) MAP kinase antibody kits (Cell Signalling Technology, Danvers, MA, USA) following the manufacturer's instructions. An α-tubulin monoclonal antibody (Sigma Aldrich) was used for the loading control.

Growth Rate as a Feature to Identify Monokaryons
We studied the growth rates on malt extract solid medium (MESM) of a progeny of 60 monokaryons obtained in 2016 from the dikaryotic strain P. ostreatus DkN001. The growth rate of each monokaryon (Mk) was the average of three biological replicas. The growth rates ranged from 0.56 to 3.08 mm/day ( Figure S1) and were highly different from those of the monokaryotic progeny derived by meiosis from the same DkN001 in 1998 (1.02 mm/day and 4.2 mm/day, data not shown). The growth rate values of monokaryons of this new progeny showed no normal distribution (Figure 1), and the nonparametric Kruskal-Wallis test was used to group monokaryons according to their growth rates. As described previously, two subpopulations of nine monokaryons (F fast and S slow) were established (Table S2).
The hyphal width and the distance between septa in the mycelia of fast and slow monokaryons were analyzed. No significant diameter differences between fast and slow monokaryons were observed (5.51 ± 1.19 µm versus 5.1 ± 1.2 µm), whereas striking differences were found in the distance between septa in fast-growing (43.56 ± 5.25 µm) versus the slow-growing monokaryons (23.41 ± 2.01 µm) ( Figure S2). The hyphal width and the distance between septa in the mycelia of fast a monokaryons were analyzed. No significant diameter differences between fast a monokaryons were observed (5.51 ± 1.19 μm versus 5.1 ± 1.2 μm), whereas strikin ences were found in the distance between septa in fast-growing (43.56 ± 5.25 μm the slow-growing monokaryons (23.41 ± 2.01 μm) ( Figure S2).

Fast-Growing Monokaryons Are Higher Biomass Producers
The phenotype differences observed between both subpopulations prompt design a whole-genome transcriptome analysis of monokaryons of both subpop to uncover their molecular bases. Because the cultures made on a solid media did enough biomass for the transcriptome analysis, each monokaryon was cultured for seven days. The Spearman coefficient was used to assess the correlation growth rate and biomass production, and the data obtained revealed a positive co value of ρ = 0.7 (p-value < 0.005) between both variables. Monokaryons of both s lations showed a different growth behavior: those of the F-subpopulation develo ial mycelia while those of S-subpopulation showed partly floating or complet merged mycelia.

RNA Sequence Analysis Allows Detection of SNPs and Crossover Locations and Re Differences between F-and S-Subpopulations
The whole transcriptome was analyzed in protoclones MkPC9 and MkPC fast-and nine slow-growing monokaryons of the progeny (F-and S-subpopulati RNA-seq analysis uncovered SNPs (Single Nucleotide Polymorphisms) MkPC9 and MkPC15, which were detected in the progeny ( Figure 2). SNP dat 190,000) allowed: (i) the identification of the parental origin of each allele, (ii) the c tion of an array of SNPs in both subpopulations, and (iii) the location of crosso regions ( Figure S3) in each chromosome of the P. ostreatus parental strain DkN00 differences were observed in 7381 out of the 11,820 genes of the whole genome treatus (transposons not considered).

Fast-Growing Monokaryons Are Higher Biomass Producers
The phenotype differences observed between both subpopulations prompted us to design a whole-genome transcriptome analysis of monokaryons of both subpopulations to uncover their molecular bases. Because the cultures made on a solid media did not yield enough biomass for the transcriptome analysis, each monokaryon was cultured in SmF for seven days. The Spearman coefficient was used to assess the correlation between growth rate and biomass production, and the data obtained revealed a positive correlation value of ρ = 0.7 (p-value < 0.005) between both variables. Monokaryons of both subpopulations showed a different growth behavior: those of the F-subpopulation developed aerial mycelia while those of S-subpopulation showed partly floating or completely submerged mycelia.

RNA Sequence Analysis Allows Detection of SNPs and Crossover Locations and Reveals Differences between F-and S-Subpopulations
The whole transcriptome was analyzed in protoclones MkPC9 and MkPC15, nine fastand nine slow-growing monokaryons of the progeny (F-and S-subpopulations).
RNA-seq analysis uncovered SNPs (Single Nucleotide Polymorphisms) between MkPC9 and MkPC15, which were detected in the progeny ( Figure 2). SNP data (about 190,000) allowed: (i) the identification of the parental origin of each allele, (ii) the construction of an array of SNPs in both subpopulations, and (iii) the location of crossover (CO) regions ( Figure S3) in each chromosome of the P. ostreatus parental strain DkN001. Allele differences were observed in 7381 out of the 11,820 genes of the whole genome of P. ostreatus (transposons not considered).
Ninety-two crossovers were distributed along the 11 chromosomes of P. ostreatus DkN001. Of the genome regions spanning crossovers, 28% (26/92 crossovers) were enriched with transposable elements (Table S3) such as gypsy retroelements and LTRretrotransposons. No correlation was observed between the crossover number and the chromosome length. Chromosomes V and XI are the fourth-and the sixth-longest chromosomes of P. ostreatus MkPC15 and showed the highest percentage of crossover (≈12%). The mating-type carrying chromosomes III and IX showed 9.78% of crossovers (Table S4). Other chromosomes with a similar crossover abundance were chromosomes VI and VII, which are highly polymorphic. A high frequency of crossovers was observed at the sub-telomeric regions of chromosomes I, II, VII, IX, and X (data no shown). SNPs analysis showed that Mk83 was achiasmatic in five of the 11 chromosomes (III, VI, VIII, IX, and XI), four of them were of the MkPC9 type. Ninety-two crossovers were distributed along the 11 chromosomes of P. ostreatus DkN001. Of the genome regions spanning crossovers, 28% (26/92 crossovers) were enriched with transposable elements (Table S3) such as gypsy retroelements and LTR-retrotransposons. No correlation was observed between the crossover number and the chromosome length. Chromosomes V and XI are the fourth-and the sixth-longest chromosomes of P. ostreatus MkPC15 and showed the highest percentage of crossover (≈12%). The mating-type carrying chromosomes III and IX showed 9.78% of crossovers (Table S4). Other chromosomes with a similar crossover abundance were chromosomes VI and VII, which are highly polymorphic. A high frequency of crossovers was observed at the subtelomeric regions of chromosomes I, II, VII, IX, and X (data no shown). SNPs analysis showed that Mk83 was achiasmatic in five of the 11 chromosomes (III, VI, VIII, IX, and XI), four of them were of the MkPC9 type.
The Spearman correlation coefficient was used to determine if any correlation existed between gene expression (RPKMs) and monokaryon growth rates (mm/day). There were 900 genes that correlated with the growth rate in monokaryons of both subpopulations; 808 of them were Positively Correlated (PC genes, ρ > 0.65, Table S7), and 92 were Negatively Correlated (NC genes, ρ < −0.65, Table S8) with the growth rate, both with a p-value < 0.005.
KOG (Eukaryotic Orthologous Group) analysis was used to determine the categories to which DEGs and PC and NC genes belong. The results obtained showed, in the Fsubpopulation, 615/808 PC genes (76%) and 136/204 F-DEGs (66%) assigned to KOG classifications ( Figure 3, Table S9). The most representative KOG categories were translational, ribosomal structure and biogenesis ( (Tables S5 and S7).
Venn diagrams were constructed to untangle any overlapping between PC and NC genes with growth rate and DEGs in both subpopulations. There were 114 and 70 genes that overlapped between the categories that were compared ( Figure S4). GO terms were assigned to these genes by using the OmicsBox software. Comparisons between overlapping DEGs in PC and NC genes (Tables S5 and S6, and Figure S5) showed that the most representative category in both populations was Cellular Component (CC), and, within it, the function of the Integral Component of Membrane with 20 and 17 genes in the F-and S-subpopulations, respectively, although the DEGs found in each subpopulation were not the same. When comparisons were performed in the category of Biological Process (BP), functions related to protein synthesis were most prominent in the F-subpopulation.
By contrast, the most relevant function in the BP and Molecular Function (MF) categories in the S-subpopulation were transmembrane transport (examples of genes in this function are: several pleiotropic drug resistance ABC transporters, Pdr15 and MFS polyamine transporters, Tpo1), and metal ion binding (wherein genes such as an alternative oxidase Aox, a flavoprotein Aif1, ID:1113817, or a histone acetyltransferase coactivator of SAGA complex Ada2, ID:1112521, could be found).

Validation of DEGs by RT-qPCR
In order to validate RNA-seq experiments, 20 DEGs belonging to different KOG categories in both subpopulations were studied. The relative expression data provided by RT-qPCR were consistent with their genome RNA transcriptome profile detected by RNA-seq and confirmed the upregulation of the analyzed genes ( Figure 4).   Table S1. Lower-case letters indicate statistically significant expression differences between F-and S-subpopulations (Student t-test, p-value < 0.005).

Gene Expression Profiles Explain Correlations between Genotype and Phenotype in Monokaryons Belonging to F-and S-Subpopulations
To investigate the genetic bases of the differences in growth rate between the F and S monokaryons, we focused on the Aox, Sod1, Cyc7, and Cox5B genes because they are related with oxidative stress and hypoxia. The two parental alleles of the Aox gene showed a 99% of protein sequence identity, though they differed in their promoters (data not shown). No differences (100% of identity) were observed between the alleles of the Sod1 gene. Both genes, Aox and Sod1, mapped to chromosome VII ( Figure 2). The Cyc7 and Cox5B genes mapped to chromosome VIII as a QTL related to the growth rate in monokaryons. No differences were found between the alleles of parental protoclones of both genes (100% identity, data not shown). The RNA-seq experiments revealed significant expression differences (detected by RPKMs, Table 1) in the four genes depending on the parental type allele they received. Other genes overexpressed in hypoxia conditions in slow-growing monokaryons of P. ostreatus were Ole1, Erg11, and Fet3 (a ferroxidase ID: 1094975) (Table S6). Table 1. Expression profile (RPKM) and parental origin allele of Sod1, Aox, Cyc7, and Cox5B genes. RPKMs were retrieved from RNA-seq data. MkPC9 (dark red) and MkPC15 (dark blue) are the parental protoclones. Light red (MkPC9 type) and blue (MkPC15 type) colors in fast-and slow-growing monokaryons indicate the origin of the parental allele in each monokaryon. Monokaryon growth rates ranged from the fastest to the slowest (indicated by the symbols + and -, respectively).

Gene Expression Profiles Explain Correlations between Genotype and Phenotype in Monokaryons Belonging to F-and S-Subpopulations
To investigate the genetic bases of the differences in growth rate between the F and S monokaryons, we focused on the Aox, Sod1, Cyc7, and Cox5B genes because they are related with oxidative stress and hypoxia. The two parental alleles of the Aox gene showed a 99% of protein sequence identity, though they differed in their promoters (data not shown). No differences (100% of identity) were observed between the alleles of the Sod1 gene. Both genes, Aox and Sod1, mapped to chromosome VII (Figure 2). The Cyc7 and Cox5B genes mapped to chromosome VIII as a QTL related to the growth rate in monokaryons. No differences were found between the alleles of parental protoclones of both genes (100% identity, data not shown). The RNA-seq experiments revealed significant expression differences (detected by RPKMs, Table 1) in the four genes depending on the parental type allele they received. Other genes overexpressed in hypoxia conditions in slow-growing monokaryons of P. ostreatus were Ole1, Erg11, and Fet3 (a ferroxidase ID: 1094975) (Table S6). Table 1. Expression profile (RPKM) and parental origin allele of Sod1, Aox, Cyc7, and Cox5B genes. RPKMs were retrieved from RNA-seq data. MkPC9 (dark red) and MkPC15 (dark blue) are the parental protoclones. Light red (MkPC9 type) and blue (MkPC15 type) colors in fast-and slow-growing monokaryons indicate the origin of the parental allele in each monokaryon. Monokaryon growth rates ranged from the fastest to the slowest (indicated by the symbols + and -, respectively).

In Situ Detection of Superoxide Radicals and Hydrogen Peroxide in Monokaryons Suggest Oxidative Stress
Due to hypoxia genes appearing upregulated in the monokaryons of the S-subpopulation, we investigated ROS production in the F-and S-subpopulations. The occurrence of superoxide radicals (O2 − ) was confirmed by the emergence of an intense dark-blue colored water-insoluble precipitate ring (formazan) of variable width, mainly on the mycelia tips of the S-subpopulation. The exceptions were Mk61 (it must be noted that this monokaryon produces a mainly floating biomass in SmF), which harbored Aox, Sod1, and Cyc7 inherited from MkPC9; and Mk75, which also ferried the MkPC9 allele type of the Cox5B gene

In Situ Detection of Superoxide Radicals and Hydrogen Peroxide in Monokaryons Suggest Oxidative Stress
Due to hypoxia genes appearing upregulated in the monokaryons of the S-subpopulation, we investigated ROS production in the F-and S-subpopulations. The occurrence of su-peroxide radicals (O 2 − ) was confirmed by the emergence of an intense dark-blue colored water-insoluble precipitate ring (formazan) of variable width, mainly on the mycelia tips of the S-subpopulation. The exceptions were Mk61 (it must be noted that this monokaryon produces a mainly floating biomass in SmF), which harbored Aox, Sod1, and Cyc7 inherited from MkPC9; and Mk75, which also ferried the MkPC9 allele type of the Cox5B gene ( Table 1). The color pattern production indicated a high accumulation of O 2 − as observed in Mk13 and Mk20 of the F-subpopulation, both bearing alleles of the MkPC9 type. Mycelial sectors were another distinctive feature in most slow-growing monokaryons and some fast-growing ones ( Figure S7A). H 2 O 2 production detected that Mk06 and Mk13 of the F-subpopulation secreted a halo of H 2 O 2 into the culture media ( Figure S7B). In addition, Mk13 and Mk20 also showed a dark central area indicating an accumulation of H 2 O 2 radicals. Intensely colored mycelia appeared in most of the slow-growing monokaryons due to the in situ activity of the Sod1 enzyme. The upregulation of the Sod1 gene was previously detected by RNA-seq experiments ( Table 1).

Assessment of Lipid Peroxidation by MDA Levels in Monokaryons of F-and S-Subpopulations
To get evidence of oxidative stress, we analyzed the lipid peroxidation of cell membranes as a biomarker in mycelia of the F-and S-subpopulations by measuring the MDA equivalents, and we found that the average MDA equivalents were 6.1 ± 1.4 and 3.7 ± 0.46 nmol/g FW for the F-and S-subpopulations ( Figure S8). A significant decrease (p-value < 0.001) of MDA levels was detected in monokaryons of the S-subpopulation. Furthermore, in slow-growing monokaryons of P. ostreatus, the Ole1 and a cytochrome b5 reductase (Cyb5R, ID: 1102329) genes involved in these processes were overexpressed (Table S6).

Monokaryons of the F-Subpopulation Are More Sensitive to Chitosan
We studied the effect of chitosan on the growth rate of 18 monokaryons grown on MESM amended with different chitosan concentrations. Analyzing the growth rate in the monokaryons of both subpopulations, we could establish four different patterns: A to D ( Figure 5).
The monokaryons displaying the chitosan response pattern A in both subpopulations showed significant growth rate differences (p-value < 0.05) at all chitosan concentrations ( Figure S9). The situation differed in monokaryons with pattern B, which showed no differences in the mycelial growth rate at 1 and 2 mg/mL chitosan, while significant differences (p-value < 0.05) were detected between the highest concentrations in comparison with the other ones. The monokaryons of both subpopulations displaying the pattern C showed no effect of chitosan on the growth rate at three chitosan concentrations, but differences were observed between them and the control sample. Only one monokaryon, Mk74, was assigned to pattern D. This monokaryon showed no reduction in growth rate when the control and the 0.5 mg/mL of chitosan samples were compared, but showed significant differences (p-value < 0.05) in comparison to the control at the highest chitosan concentrations ( Figure S9).

Monokaryons of the F-Subpopulation Are More Sensitive to Chitosan
We studied the effect of chitosan on the growth rate of 18 monokaryons grown on MESM amended with different chitosan concentrations. Analyzing the growth rate in the monokaryons of both subpopulations, we could establish four different patterns: A to D ( Figure 5). The monokaryons displaying the chitosan response pattern A in both subpopulations showed significant growth rate differences (p-value < 0.05) at all chitosan concentrations

Effect of Chitosan Treatments on Some Genes Upregulated in Monokaryons of the F-and S-Subpopulations
RNA-seq experiments performed in this work showed that two chitin deacetylase Cda1 genes were overexpressed in slow-growing monokaryons, suggesting they are chitosan producers. To determine how this production could affect growth rate and promote the expression of genes involved in oxidative stress and cellular detoxification, such as Aox, Sod1, and Pdr15, we analyzed the effect of different chitosan concentrations on the F-and S-subpopulations by assessing the expression profile of the genes mentioned above.
The expression profile of the Aox gene was analyzed in two fast (Mk01 and Mk20) and two slow (MK69 and Mk83) growing monokaryons classified in the A and C patterns using RT-qPCR. The RPKM expression data were estimated as described in the Figure 6 legend. Compared to the control sample, the fast-growing monokaryons increased their relative expression values, mainly at 0.5 and 1 mg/mL of chitosan. The highest (2 mg/mL) chitosan dose induced a limited expression increase in Mk01 but was ineffective in Mk20 ( Figure 6A). A substantial growth rate reduction was observed upon chitosan addition ( Figure S9) in both monokaryons. The relative expression of the Aox gene in slow-growing monokaryons revealed quite different landscapes. In Mk69, chitosan Aox induction was increased at high doses, while in Mk83, the highest response was obtained at 1 mg/mL ( Figure 6A). It should be mentioned that the two slow-growing monokaryons showed a similar growth rate pattern ( Figure S9) but different parental origins of the Aox alle-les ( Table 1). The estimated RPKMs of the Aox gene showed that the control sample of Mk83 displayed the highest value compared to the other control samples of the rest of the monokaryons ( Figure 6B).
The expression analysis of the Sod1 gene followed a relatively similar expression profile to the Aox gene ( Figure 6C): the individuals with highest Aox expression had the highest Sod1 RPKM values ( Figure 6D). As both genes are placed at chromosome VII, and no crossover was detected, it could be said that both genes were inherited as a supergene.
The expression of the gene Pdr15 was higher at lower chitosan concentrations, and at the highest concentration of this antifungal, the gene showed a mild effect on Mk01 and no effect on Mk20 ( Figure 6E). The relative expression of Pdr15 in slow-growing monokaryons displayed a similar profile to that observed in the Sod1 gene. The estimation of RPKMs of the Pdr15 gene ( Figure 6F) was also analyzed. We observed that the control samples of the slow-growing monokaryons exhibited the highest RPKM values compared to the control samples of the fast-growing ones.

Analysis of Mitogen-Activated Protein Kinase (MAPK) Signalling Pathways in Monokaryons of the F-and S-Subpopulations
We have studied the Mitogen-Activated Protein Kinase (MAPK) pathways in monokaryons of the F (Mk01, Mk08, and Mk20) and S (Mk61, Mk69, and Mk83) subpopulations. Total protein extracts of monokaryons were immunoblotted and revealed using commercial antibodies against the unphosphorylated and phosphorylated forms of p38 (HOG) and p44/42 (ERK1/2) of CWI pathways to measure the endogenous levels of both kinases in monokaryon protein extracts (Figure 7).   Twenty five micrograms of total protein extract from three fast (Mk01, Mk08, and Mk20) and three slow (Mk61, Mk69, and Mk83) growing monokaryons cultivated in SMY liquid medium for seven days were subjected to immunoblot analysis with an anti-phospho-p38 MAPK antibody and an anti-phospho-p44/42 MAPK antibody to detect phosphorylation of these MAPKs. A monoclonal α-tubulin antibody was used as a loading control. The intensity of the phosphorylation signal is inversely correlated with the growth rate of mycelia.
The unphosphorylated p38 (45 kDa) band showed a similar intensity in Mk20 (the slowest of the F-subpopulation) and the monokaryons of the S-subpopulation Mk61 and Mk69. When the phosphorylated version of p38 (Phospho p38), which recognizes the TGY motif characteristic of stress-activated MAPK activated by phosphorylation in threonine (T) and tyrosine (Y), was probed, two bands of 45 (phosphorylated p38) and 52 (unknown origin) kDa were detected. The 45 kDa band lighted up in two of the three fast-growing and the three slow-growing monokaryons.
We also analyzed the CWI pathway in fast-and slow-growing monokaryons. The antibodies against p44/42 detected a predominant band of 42 kDa with a similar intensity in all samples except in the fast-growing Mk83 that was fainter. In contrast, the antibody against phosphorylated p44/42 showed a signal in the slow-growing monokaryons and the fast-growing Mk20. We analyzed if any correlation existed between the strength intensity signal of the bands and RPKM values of the Cda1 genes (data not shown). The results showed that the high activation level of the p44/42 protein corresponded with the highest RPKM values of the Cda1 genes of the slow-growing monokaryons, where these genes were overexpressed, and chitosan was produced.

Discussions
A global discussion of all of the topics developed in this paper is presented. Firstly, we focus on the mycelial growth rate and cell size in fast-and slow-growing monokaryotic subpopulation progenies obtained from the dikaryotic strain DkN001 in 2016, compared with the same data recorded from similar subpopulations obtained from the same dikaryon in 1998 [68]. In fact, we have seen that the progeny obtained in 2016 showed a drastic decrease in mycelia growth rate compared to that of 1998 [52]. In this sense, the microscope analysis of mycelial hyphae of F-and S-subpopulations described in this paper ( Figure S2) showed that both had a similar diameter. However, increased septation was only present in hyphae of monokaryons belonging to the S-subpopulation. Chiu et al. [69] described that yeast cells subjected to some oxidative stresses delay entry at the G1 phase of the cell cycle, allowing them to repair any cellular damage. This observation means that oxidative stress sensing would be coordinated with the regulation of the cell cycle. The increase in the number of septa observed in monokaryons of the S-subpopulation is compatible with a curtailment of the cell cycle, probably due to endogenous changes, leading to metabolic switching.
Differences between the two subpopulations of monokaryons uncovered by the combined analysis of the whole-genome transcriptome and SNPs revealed: (i) that the different phenotypic features observed in monokaryons of both subpopulations were based on allele differences and (ii) that genetic recombination in P. ostreatus was restricted to telomeric and subtelomeric regions, as has been reported in other basidiomycetes [70]. However, in Pleurotus, a lack of CO in some chromosomes was observed. This last fact assures that a QTL for monokaryotic growth rate placed in chromosome VIII, apart from some other genes, was inherited as a supergene, and it was a clear-cut example of a low genome variability observed in the progeny of P. ostreatus.
Data analysis of the whole-genome transcriptome profile of both subpopulations yielded different landscapes with genes overexpressed in the S-subpopulation that inform about significant metabolic changes. Although the traditional loss-of-function analysis has been a straightforward mechanism to uncover a gene function, the overexpression analysis is an alternative tool to identify pathway components that may remain undetected using a knock-out approach [71], even more, when overexpression mirrored the physiology of the slow-growing monokaryons, as is shown in this paper, We jointly analyzed DEGs (in this paper, overexpressed genes) and genes PC and NC with the growth rate in both subpopulations and found that the transcriptome profile yielded 900 genes correlated with growth rate and 531 DEGs (as was described in Section 3.4). About 73% of transcripts were genes PC with growth rate in the Fsubpopulation, but only 38.4% were F-DEGs. Most of them belonged to cellular processing and signaling (KOGs: O, U, T, and Z) and information storage and processing (KOGs: J, A, and K) (Table S9), which indicated that monokaryons of the F-subpopulation showed signals of active growth under the experimental conditions described in this work. By contrast, monokaryons of the S-subpopulation showed only 10.2% of genes were NC with growth rate, and about 61.6% were S-DEGs. It should be noticed that 11.8% of S-DEGs were located in chromosome VIII, and most of them were related to the metabolism and transport of different molecules. Genes found in the S-subpopulation, mainly oxidative stress genes, undoubtedly indicated that other metabolic conditions, different from those found in fast-growing monokaryons, have been developed under identical culture conditions, which have led to symptoms of strain degeneration. Suparmin et al. [72] obtained similar results under identical experimental conditions, though the metabolic changes observed in the monokaryons of F-and S-subpopulations shed light on the different genetic backgrounds of both subpopulations. Overlapping analysis performed between F-DEGs and PC genes versus S-DEGs and NC genes using Gene Ontology (GO) ( Figure S4) reinforced, for each subpopulation, the data obtained by KOG analysis.
A combined analysis of MkPC9 and MkPC15 SNPs and RNA-seq experiments carried out in the parental protoclones and in the progeny of the F-and S-subpopulations shed light on the performance of the monokaryons growth rate in liquid and solid media. This clue was provided by the analysis of RNA-seq experiments of four genes and growth rate QTLs. The genes analyzed were: (i) Aox coding for a non-energy-conserving terminal oxidase of the Electron Transport Chain (ETC) with a lower affinity for oxygen than cytochrome oxidase, and overexpressed under stress conditions [73,74]; (ii) Sod1 encoding a superoxide dismutase which regulates the production of ROS and is required for tolerance to oxidative stress [75,76]; (iii) Cyc7 (isoform 2 of cytochrome c, ID:1113744) that is an electron carrier protein predominant under anaerobic/hypoxic conditions; (iv) and Cox5B (cytochrome c oxidase subunit 5B, ID:1094413) that is a subunit of the ETC Complex IV expressed under hypoxic conditions [77].
We observed that expression levels of the Aox and Sod1 genes in the progeny depended on the parental allele inherited. The high levels of the Aox gene in the S-subpopulation suggested an impairment or not a fully functional respiratory chain, which could increase intracellular ROS. These high ROS-induced Aox levels could be compensated with high induction levels of the Sod1 gene, the overexpression of which decreased cellular susceptibility to oxidative stress by promoting cell adaptation as described by Li et al. [78] in cancer cells. Both genes, cosegregated in the progeny and their expression and the in situ ROS detection in monokaryons allowed us: (i) to differentiate both subpopulations, (ii) to corroborate the intense oxidative stress supported by monokaryons of the S-subpopulation, and (iii) to show the link between the genotype and the phenotype performance observed in the progeny. This link was also observed when the hypoxia-induced genes Cyc7 and Cox5B were studied. Kwast et al. [77] observed that mitochondrial cytochrome c oxidase (COX) subunit 5 and cytochrome c (Cyc) existed in two isoforms in yeast, which were transcriptionally regulated by oxygen. The gene pair Cox5A/Cyc1 encodes the normoxic isoforms (Cox5A and iso1-Cyc), while the gene pair Cox5B/Cyc7 encodes the hypoxic isoforms (Cox5B and iso2-Cyc). Castello et al. [79] described that the hypoxic genes Cyc7 and Ole1 were induced at higher levels in yeast strains carrying the Cox5B isozyme. Rox1 is a transcriptional repressor of Cox5B and Cyc7 genes in normoxia. In P. ostreatus, the expression of the Rox1 homolog gene (ID: 1017219, Table S7) was positively correlated with growth rate, which means that its downregulation in monokaryons of the S-subpopulation led to high expression of the hypoxic isoforms Cox5B and Cyc7.
As no differences between the parental alleles (100% of protein sequence identity) of the loci Cox5B/Cyc7 were found, the differences in the expression levels between fast-and slow-growing monokaryons could depend on Rox1 gene expression. In fact, it could be accepted that the expression of Rox1-regulated hypoxic genes would increase when oxygen concentrations decrease. Liu et al. [80] showed that, in yeast, oxidative stress induced the expression of the OXPHOS-related Cox5B and Cyc7 genes through a mechanism that released the transcriptional repressor Rox1 from their promoters. All these data show that monokaryons of the S-subpopulation showing strain degeneration are under intense oxidative stress.
The imbalance of ROS leads to an excess of radical species that produce adverse modifications to cell components such as lipids, proteins, and DNA damage [81]. As we detected by a proper staining method, the activity of the Sod1 enzyme, especially in slow-growing monokaryons, we assessed the malondialdehyde levels (MDA, a biomarker of oxidative stress depending on the composition of saturated/unsaturated fatty acids present in cell membranes, the substrate for free radicals) resulting from peroxidation of biological membranes in fast-and slow-growing monokaryons [67]. Lipid peroxidation is a free radical-mediated chain of reactions that, once initiated, results in an oxidative deterioration of polyunsaturated fatty acids (PUFA) [82]. High levels of MDA correspond to cell membranes containing a high percentage of PUFA, while low levels of MDA correspond to membranes with a high percentage of Saturated Fatty Acids (SFA).
Due to their high PUFAs, cellular membranes or organelle membranes are especially susceptible to ROS damage producing lipid peroxidation. This process, in which free radical species (such as oxyl, peroxyl, and hydroxyl radicals) remove electrons from lipids, produce reactive intermediates that can undergo further reactions [84]. The lower MDA levels observed in the S-subpopulation monokaryons than those obtained in the Fsubpopulation could suggest a higher membrane strength supported by the overexpression of the hypoxic Cyc7, Cox5B, and Ole1 genes. The overexpression of the Cyc7 and Cox5B genes would indicate that oxygen availability plays a crucial role in the composition of cell membranes.
The Ole1 gene encodes a stearoyl ∆ 9 desaturase regulated at transcriptional (by Rox1 [77]) and post-transcriptional levels by different factors such as glucose, fructose, cholesterol, PUFAs, and senescence, among others. This enzyme catalyzes the conversion of SFA to MUFA. Different functions have been assigned to this enzyme. Still, the most important one is to maintain the fatty acid composition and membrane structure because high levels of SCD could lead to rigid membrane composition (low membrane fluidity), apoptosis resistance [83,85], and extension of lifespan as was observed in monokaryons of the S-subpopulation. Nasution et al. [86] described that Ole1 overexpression might enhance tolerance to various types of stress because it would produce an accumulation of MUFA, which determines a lower membrane permeability. It was observed that cell membranes of some entomopathogenic fungi belonging to the Cordycipitaceae family and nematophagous egg-parasitic fungi of the Clavicipitaceae families changed their membrane fluidity to reduce chitosan damage [40].
Chitosan was described as a powerful biocidal [87], affecting the mycelial growth rate and the expression of genes related to cell detoxification and hypoxia. The effect of chitosan on growth rate was studied in monokaryons of the F-and S-subpopulations, and the results obtained indicated that the decrease in growth rate was genotype dependent. In this sense, most fast-growing monokaryons ferried the growth rate QTL of the MkPC9 type and the Cox5B and Cyc7 allele expression compatible with normoxia conditions. Conversely, slow-growing monokaryons harbored the growth rate QTL of the MkPC15 type, the Cox5B and Cyc7 alleles expressed under hypoxia conditions, and displayed mycelial degeneration (Table 1 and Figure 2).
Palma-Guerrero et al. [40] related fungal chitosan sensitivity to high levels of PUFA and increased plasma membrane fluidity. In the light of that result, we speculated that monokaryons of the F-subpopulation could have a higher level of PUFAs than those of the S-subpopulation because their growth rate was reduced in the presence of chitosan, and their MDA levels were higher than those observed in slow-growing monokaryons overexpressing the Ole1 gene.
Because chitosan induces membrane permeation and increases the production of intracellular ROS [28], the response of monokaryons of the F-and S-subpopulations to different chitosan concentrations was studied by analyzing the expression levels of the Aox and Sod1 genes. The high expression of the Aox gene at 2 mg/mL of chitosan in Mk69 could be explained as an adaptive response of the monokaryotic mycelium grown for 10 days under different chitosan concentrations. Thus, these culture conditions would lead to a stimulation effect of the alternative respiration due to increasing Aox transcripts. By contrast, constitutive expression of Aox and Sod1 genes in the control sample of Mk83, which inherited MkPC15 type alleles, could fit with a monokaryon already adapted to chitosan (resistant to chitosan). We hypothesized that Aox would sense the ROS production, and Sod1 would stimulate a chitosan defense/adaptive reaction due to both genes showing an increase of their transcripts in those monokaryons with inherited MkPC9 type alleles. Knock-out experiments alternatively in both genes could provide some additional clues.
Many stress response genes were overexpressed in the whole transcriptome profile of monokaryons of the S-subpopulation. One of them was the Ada2 gene, a histone acetyltransferase coactivator of the SAGA complex, which can bind to about 200 gene promoters including those related to glycolysis, pyruvate metabolism, oxidative stress, drug responses, cell wall response [88], and genes involved in ergosterol biosynthesis [89].
This paper focused on the Pdr15 gene, which was described as the first pleiotropic drug-resistant ABC transporter involved in a general stress response [90] in monokaryons of the F-and S-subpopulations. The expression of this gene was increased in fast-growing monokaryons, probably due to the ROS insult produced by chitosan. The increase of relative expression of this detoxification gene could indicate how fast-growing monokaryons fight off different chitosan concentrations. By contrast, the behavior showed by the slowgrowing monokaryons could be understood as the result of an adaptive response/resistance mechanism developed to cope with different chitosan concentrations. This finding could suggest that the Pdr15 gene, as the other genes mentioned above, were constitutively expressed in the S-subpopulation to alleviate ROS and altogether could constitute markers of strain degenerated mycelia.
Due to the RT-qPCR experiments being carried out in the stationary phase (ten-day culture under chitosan treatment), we assumed that nutrient availability was insufficient, as shown by the overexpression of the hexose transporter Rco-3 (ID:1094871, Table S10). Thus, we supposed that toxic metabolites and catabolites had started to accumulate, yielding cellular damage and the overexpression of some apoptotic genes, such as an aspartic protease Ap1 (ID:1033895), an apoptosis-inducing factor Aif1 (ID:1113817), and a metacaspase Mca1 (ID:1114660) (Table S10); despite the overexpression of the Ole1 gene, which is involved in the extension of lifespan [83]. Thus, in slow-growing monokaryons, we suggest that Pdr15 could be induced to alleviate oxidative stress as described in yeast [90], and together with Aox, Sod1, and Ole1 genes, could be used to distinguish mycelia of strain degenerated strains.
Other genes upregulated by the Ada2 gene, the overexpression of which was independent of chitosan addition [88], were observed in slow-growing monokaryons. The list encompasses genes such as (i) Erg1 (squalene synthase) and Erg9 (squalene epoxidase) (Table S10), which are considered sterol precursors involved in the ergosterol biosynthesis pathway and their overexpression is the result of a depletion of cellular ergosterol in hypoxic conditions [91], (ii) the Prz1 transcription factor, a Schizosaccharomyces homolog to Crz1 [92], regulated by the Ca 2 + /calmodulin-dependent protein phosphatase (calcineurin) and activated under certain environmental conditions, and (iii) the Pho84 gene, a pleiotropic high-affinity inorganic phosphate transporter the overexpression of which triggers the Endoplasmic Reticulum Unfolded Protein Response (ER-UPR) related to the extension of Replicative LifeSpan (RLS) in the retrograde response in yeast [12].
Changes in environmental conditions demand quick shifts in cell wall restructuring via the synthesis of chitin and chitosan enzymes. In this work, we have seen that monokaryons of the S-subpopulation overexpressed the Uap1 and two Cda1 genes, both involved in cell wall remodeling [93]. As previously described, the Uap1 gene synthesizes chitin monomers, while Cda1 genes code for the enzymes necessary to deacetylate either the chitin polymer or the chitin monomer resulting from the activity of the Uap1 gene with the aim of reshaping the cell wall. In Cryptococcus [94], the Cda genes are expressed during vegetative growth to provide integrity and proper rigidity to the cell wall and cope with the effect of self-chitosan production. Recently, Rizzi et al. [95] described Cda genes as necessary for cell wall integrity and viability.
It was described that the contact between Trichoderma, a mycoparasitic fungus, with its prey triggered a burst of oxidative stress detected by an increase of expression of chitin synthase and chitin deacetylase genes [96]. P. ostreatus is a basidiomycete with the predatory lifestyle of a nematode-trapping fungus. This lifestyle has evolved among cellulolytic or ligninolytic fungi due to nutrient deficiencies in nitrogen-limiting habitats [97]. We rule out that the increase of Uap1 and Cda1 genes were due to contact with any prey because experiments were performed under controlled conditions. Thus, we speculate that the overexpression of Uap1 and Cda1 genes in slow-growing monokaryons of P. ostreatus, with symptoms of strain degeneration, could be accounted for: (i) as a self-defense mechanism acting as a response to a signal molecule, chitosan, with the purpose to strengthen cellular membranes and restructure cell walls. Remodeling of cell walls could permit it to stand harsh oxidative stress conditions which could lead to an extension of lifespan and could be assumed due to the overexpression of the Pho84 gene, and (ii) due to nutrient depletion that takes place during the stationary phase when experiments were performed, which would trigger the recycling of the cell wall monomers.
Alternatively, strain degeneration in slow-growing monokaryons could result from the impairment between nuclear and mitochondrial genes related to the functioning of the ETC. In P. ostreatus, mating between compatible mycelia entails the migration of nuclei (highly unusual mitochondrial migration). Hence, the newly formed dikaryon and its progeny only bring a mitochondrial type. This fact could be responsible for the weak performance of slow-growing monokaryons with a long-lasting cellular cohabitation period between unmatched nuclear and mitochondrial ETC genes. Currently, this fungal aspect in P. ostreatus is under study in our laboratory.
Taken together, we conclude that strain degeneration signs observed in slow-growing monokaryons of P. ostreatus could be the result of multiple factors that determine that strain degeneration can be seen as a pleiotropic trait.
Two pathways, the High Osmolarity Glycerol (HOG) and Cell Wall Integrity (CWI), regulate responses to stress conditions [98]. The HOG pathway is activated by different stresses [99,100], while the CWI pathway is necessary for the integrity of the cell wall [101]. The first one, once activated, coordinates the hyperosmotic stress response, though it could also be activated in response to other stresses, including oxidative [102], acid [103], methylglyoxal [104], temperature downshift [105], or heat stresses [106]. The CWI pathway is mainly activated by cell wall stress to ensure cell wall integrity by increasing chitin synthesis [96]. However, this pathway can be activated by some other stressful conditions, including hypo-osmotic shock [107], pheromone treatment [108], actin depolarization [109], the Unfolded Protein Response (UPR) [110], and oxidative stress, suggesting that although the pathway specificity is necessary, the HOG and CWI pathways can also be positively coordinated [98].
Immunoblots performed in mycelia of slow-growing monokaryons in the absence of chitosan showed the activation of the HOG pathway. Due to slow-growing monokaryons being themselves chitosan producers, it could be suggested that HOG activation is required to counteract the membrane insult resulting from this production in the slow-growing monokaryons and two of the three fast-growing monokaryons. Crosstalk between the stress response and the pleiotropic drug resistance Pdr15p was observed by Wolfgert et al. [90] because its induction bypasses upstream components of the HOG pathway. The overexpression of up to two Pdr15 genes in slow-growing monokaryons (Table S10) revealed by RNA-seq experiments could suggest that its overexpression could also help HOG pathway activation.
Chitin synthesis and chitosan conversion are strongly influenced by external parameters, such as environmental stress factors. Perturbation of fungal cell wall synthesis triggers a compensatory response to ensure CWI by increasing chitin synthesis [96]. We assume that a perturbation of the fungal cell wall occurring in the slow-growing monokaryons and the fast-growing Mk20 accounted for the constitutive overexpression of the Uap1 gene responsible for the synthesis of chitin monomers. This new synthesis, which would indicate cell wall restructuring, could activate the CWI pathway.
In summary, the results obtained provide evidence that (i) strain degeneration in P. ostreatus is linked to oxidative stress, (ii) the oxidative stress response in monokaryons of the F-and S-subpopulations is genotype dependent, (iii) marker genes of oxidative stress and cellular detoxification are highly expressed in monokaryons of the S-subpopulation with symptoms of strain degeneration, (iv) chitosan addition to monokaryons of both subpopulations uncovered the constitutive expression of Aox, Sod1 and Pdr15 genes in slow-growing monokaryons strains suggesting their adaptation to oxidative stress, and (v) the overexpression of the cell wall genes, Uap1 and Cda1, in slow-growing monokaryons with a strain degeneration phenotype indicated cell wall reshaping and the activation of HOG and CWI pathways. Thus, these results could constitute a hallmark for mushroom producers to distinguish strain degeneration in commercial mushrooms.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available because other data from these wholegenome transcriptomes are being used for other analyses to be published independently of this one.

Conflicts of Interest:
The authors declare no conflict of interest.