A transcription factor, MrMsn2, in the dimorphic fungus Metarhizium rileyi is essential for dimorphism transition, aggravated pigmentation, conidiation and microsclerotia formation

Summary Microsclerotia (MS) are pseudoparenchymatous aggregations of hyphae of fungi that can be induced in liquid culture for biocontrol applications. Previously, we determined that the high‐osmolarity glycerol (HOG) signalling pathway was involved in regulating MS development in the dimorphic insect pathogen Metarhizium rileyi. To further investigate the mechanisms by which the signalling pathway is regulated, we characterized the transcriptional factor MrMsn2, a homologue of the yeast C2H2 transcriptional factor Msn2, which is predicted to function downstream of the HOG pathway in M. rileyi. Compared with wild‐type and complemented strains, disruption of MrMsn2 increased the yeast‐to‐hypha transition rate, enhanced conidiation capacity and aggravated pigmentation in M. rileyi. The ▵MrMsn2 mutants were sensitive to stress, produced morphologically abnormal clones and had significantly reduced MS formation and decreased virulence levels. Digital expression profiling revealed that genes involved in antioxidation, pigment biosynthesis and ion transport and storage were regulated by MrMsn2 during conidia and MS development. Taken together, our findings confirm that MrMsn2 controlled the yeast‐to‐hypha transition, conidia and MS formation, and virulence.

Filamentous fungal Msn2/4 homologues are C 2 H 2 -like zinc finger TFs that regulate the general stress response, pathogenicity, secondary metabolism and cellular growth. They are similar to ScMsn2/4 of Saccharomyces cerevisiae (Schmitt and Mcentee, 1996) and have been characterized in Aspergillus parasiticus, A. flavus, Beauveria bassiana, Magnaporthe oryzae, Metarhizium robertsii and Verticillium dahliae (Chang et al., 2011;Liu et al., 2013;Zhang et al., 2014;Tian et al., 2017). Under abiotic and biotic stresses, Msn2/4 is phosphorylated for translocation from the cytoplasm to the nucleus, where it drives the transcription of stressinduced genes (Hansen et al., 2015;Yi and Huh, 2015;Li et al., 2017). The underlying mechanism of the regulation of Msn2/4 activity by protein kinase A (PKA), the rapamycin signalling pathway, the Snf1 protein kinase pathway and the high-osmolarity glycerol (HOG) pathway have been identified Zhang et al., 2014;Li et al., 2017).
Microsclerotia (MS) are pseudoparenchymatous aggregations of hyphae with a diameter of 50-600 lm and become melanized during their development. As promising fungal propagules, MS can be induced in liquid culture and used for biocontrol applications such as biofungicides, bioherbicides, bionematicides and mycoinsecticides (Shearer, 2007;Jackson et al., 2010;Song et al., 2014Song et al., , 2016a. To enhance the liquid fermentation efficiency of Metarhizium rileyi MS, we previously investigated the molecular mechanism of MS formation and demonstrated that internal oxidative stress triggers MS differentiation (Song et al., 2013(Song et al., , 2016b(Song et al., , 2018Jiang et al., 2014). We found that HOG and cell wall integrity (CWI) pathways cooperate to regulate MS formation (Song et al., 2016b). We also found that M. rileyi MrMsn2 was predicted to function downstream of the HOG pathway and was upregulated during MS formation in comparative transcriptome analysis (Song et al., 2013). Furthermore, a bioinformatics analysis found no Msn4 orthologues in any public genome databases of M. rileyi (Song et al., 2013;Shang et al., 2016). These results imply a possible involvement of MrMsn2 in the regulation of MS development. However, this function has not been studied clearly.
Moreover, M. rileyi is a well-known dimorphic entomopathogenic fungus with yeast-like hyphal bodies and a true filamentous growth phase (Boucias et al., 2000;Fronza et al., 2017), which occurs synchronously in vivo and in vitro (Pendland and Boucias, 1997;Boucias et al., 2016). The yeast-to-hypha transition is critical for the pathogenesis and life cycle of dimorphic fungi (Wanchoo et al., 2009;Boyce and Adrianopoulos, 2015;Gauthier, 2015;Marcos et al., 2016). Although signalling pathways related to dimorphic transition are well characterized in the model yeast Candida albicans, the mechanisms are not well defined (Noble et al., 2017). Thus, studies on M. rileyi are useful model for understanding the dimorphic transition mechanism.
This study seeks to further elucidate the role of MrMsn2 in dimorphism transition, conidiation, virulence and MS formation by phenotypic analyses of deletion/ rescue mutants constructed previously Song et al., 2016b). We found that the absence of MrMsn2 resulted in increased yeast-to-hypha transition rate, enhanced conidiation capacity, aggravated pigmentation and induced or suppressed expression of target genes involved in the important phenotypes of M. rileyi, as presented below.

Molecular characterization of MrMsn2
The full-length sequence of MrMsn2 (GenBank Accession No.: MG641237) is 1752 bp, including one intron, with a calculated molecular weight of 56.7 kDa and an isoelectric point of 5.06 (http://expasy.org/tools/protpara m.html). Furthermore, MrMsn2 contains a zinc finger double domain (Hu et al., 2014). In this study, a phylogenetic tree analysis demonstrated that the MrMsn2 protein from M. rileyi was closely related to other Metarhizium spp. proteins (Fig. S1). In addition, the amino acid sequence of MrMsn2 showed similarities (79-81% identity) to a cutinase G-box binding protein of Metarhizium spp. (Hu et al., 2014;Shang et al., 2016) and a zinc finger protein (66% identity) of Tolypocladium ophiglossoides (Quandt et al., 2015).
To characterize the functions of the MrMsn2 gene in M. rileyi, gene replacement mutants and complementary transformants were generated. All recombinant strains were verified by polymerase chain reaction (PCR) and quantitative real-time PCR (RT-qPCR) screening (Fig. S2). The confirmed MMrMsn2 mutants and the complemented (MMrMsn2 + Msn2) strains were used in further experiments.

MrMsn2 negatively regulates yeast-to-hypha transition and conidiation
The M. rileyi CQNr01 (wild-type, WT) strain was grown on solid Sabouraud maltose agar fortified with yeast extract (SMAY). Compared with the initial results at day 0, expression of MrMsn2 was found to be downregulated in the yeast-to-hypha transition at days 2 and 4 and conidiation initiation at day 6 ( Fig. 1A). These results indicate that MrMsn2 may be involved in the yeast-tohypha transition and conidiation.
Further investigations showed that at day 3, the yeast-to-hyphae transition was advanced in MMrMsn2 mutants compared to the WT and complemented (CP) strains (Fig. 1B). Furthermore, colony surfaces of MMrMsn2 mutants were more convoluted compared to the normal smooth colony surfaces of WT and CP strains. After 6 days, the diameter of the mutant colonies was larger compared to that of the WT (Fig. 1B). Additionally, the MMrMsn2 mutants had significantly increased conidial yields compared to the WT and CP strains (P < 0.001) (Fig. 1C). After 9 and 12 days, conidial yield had increased by 1.5-to 2.2-fold in MMrMsn2 mutants compared to the WT and CP strains. Taken together, these transcription and phenotype growth analysis suggest that MrMsn2 is involved in negative control of conidia production and yeast-tohypha transition.
To further analyse the effect of MrMsn2 on the dimorphic transition, yeast cells of test strains was grown on SMAY medium. The investigations into the switching rates and median transition time required for 50% transition of blastospores to hyphae (TT 50 ) showed a significantly difference among the MMrMsn2 mutants (TT 50 = 5.4 AE 0.2 days), WT (TT 50 = 7.3 AE 0.1 days) and CP (TT 50 = 7.0 AE 0.2 days) strains (P < 0.001) (Fig. 2). This suggests that deletion of MrMsn2 promoted the yeast-to-hypha transition.

Absence of MrMsn2 leads to aggravated pigmentation
After incubation on SMAY medium for 6 days, MMrMsn2 mutants were found to have altered aggravated pigmentation (Fig. 3A). After 9 and 12 days, the tergal pigment of MMrMsn2 mutants had increased compared to WT and CP strains respectively. Furthermore, the pigment concentration in MMrsn2 mutants was found to be significantly increased by factors of 1.8-to 4.2-fold, compared to that of WT and CP strains respectively (Fig. 3B). These data suggest that MrMsn2 had a negative influence on clone pigment biosynthesis.
To investigate the mechanism of dimorphic transition, pigment biosynthesis and conidiation regulated by MrMsn2, genes that were potentially involved were selected from transcriptome libraries (Song et al., 2013(Song et al., , 2018 and examined by transcriptional analysis. The following genes were selected: pigment biosynthesis-related genes (polyketide synthase, pks; polyketide synthasenon-ribosomal peptide synthetase, pks-N; polyketide synthase 6, pks-6; conidial yellow pigment biosynthesis, cyp; and laccase, lac), several chitin synthase genes (ch1, ch2 and ch4, for class I, II and IV chitin synthases respectively) and antioxidant enzyme genes (cat1 for catalase-1 and cat2 for catalase-2, sod1 for superoxide dismutase-1 and sod2 for superoxide dismutase-2). It was found that the lac, pks-N, cat1 and sod1 genes were upregulated, whereas pks, pks-6, cyp, ch1 and ch2 genes were downregulated in the WT strain during conidiation (Figs. 3C and D). Compared with the WT, all pigment biosynthesisrelated gene and chitin synthase genes were significantly upregulated during conidiation in the MMrsn2 mutants ( Fig. 3C). In addition, the antioxidant enzyme genes were upregulated in aggravated pigmentation after 6 days for MMrsn2 mutants (Fig. 3D).

MrMsn2 contributes to tolerance to abiotic stress
To examine the function of MrMsn2 on the abiotic stress response, strains were cultured under various abiotic stress conditions. Convoluted colony surfaces were more apparent, especially under cell wall perturbation and oxidative stress, for MMrMsn2 mutants compared to the normal smooth colony surfaces of the WT and CP strains, after 3 day incubation (Fig. S3A). Furthermore, compared to WT strain, smaller colonies were present in the MMrMsn2 mutants (Fig. S3A). After 12 days, the conidial yield of MMrMsn2 was significantly reduced between 26.6 and 91.9% on SMAY medium, under osmosensitivity, cell wall perturbation or oxidative stress (P < 0.001). Interestingly, the conidial yield of MMrMsn2 mutants was found to be significantly reduced under KCl stress, however, it was significantly increased under NaCl stress compared to the WT and CP strains A. Transcription of MrMsn2 during conidiation on different cultivation days. 3 ll of conidial suspensions (1 9 10 7 conidia ml À1 ) was spotted on SMAY plates and cultured under continuous light at 25°C for 8 days. Stages of conidial development: inoculated conidia at initial culture time (day 0), blastospores (day 2), hyphal period (day 4), conidiation initiation (day 6) and conidia at start of maturation (day 8). B. Images of colonies on SMAY plates for 3, 6, 9 and 12 days. 3 ll of conidial suspensions (1 9 10 7 conidia ml À1 ) was spotted on SMAY plates and cultured under continuous light at 25°C for 12 days. Bar = 5 mm.
C. Conidial yield of tested strains on SMAY medium after 6, 9 and 12 days of incubation. Error bars are standard error. * P < 0.05, ** P < 0.01, significantly different compared to wild type or 0 day incubation.

MrMsn2 is needed for MS development
An expression analysis showed that the relative transcriptional of MrMsn2 peaked with MS initiation (72 h) (Fig. 4A) and MrMsn2 were upregulated in liquid amended medium (AM) or MM (AM without basal salts) medium, cultured with exogenous oxidative stress (Fig. 4B). These results suggest that MrMsn2 may be involved in the regulation of MS formation. After incubation in liquid AM for 144 h, MS produced by WT and CP strains matured and were accompanied by secondary mycelia growth, whereas the density of the induced MS in the MMrMsn2 mutants was significantly decreased, with the MMrMsn2 culture broth exhibiting a low degree of pigmentation (Fig. 4C). Compared to the WT and CP strains, the MS yield of MMrMsn2 mutants was reduced by approximately 88.9% (Fig. 4D), and the biomass was decreased by 23.6% in the AM culture (Fig. 4E). These results indicate that MrMsn2 is needed for MS development.

Expression analysis during MS development
To investigate the genes regulated by MrMsn2 during MS formation, several groups of genes were analysed by qRT-PCR. It was found that antioxidation genes such as cat1, cat2, sod1, sod2 and the monooxygenase (mon) gene were upregulated in the MMrMsn2 mutants, while the flavoprotein-ubiquinone oxidoreductase (fuo) gene was downregulated (Fig. 5A). Pigment biosynthesis genes pks, pks-6 and cyp were found to also be downregulated and pks-N and lac genes were upregulated in the MMrMsn2 mutants (Fig. 5A). Additionally, chitin synthase genes ch2 and ch4 were significantly upregulated in the MMrMsn2 mutants and ch1 was significantly downregulated (Fig. 5B). Interestingly, Slt2, the core gene of the CWI signalling pathway, and hog1, the core gene of the HOG signalling pathway, were significantly upregulated in the MMrMsn2 mutants (Fig. 5B). Finally, transport and storage genes for major salts, sidA (siderophore iron transporter), ct-1 (calcium-transporting ATPase 1), ct-2 (calcium-transporting p-type ATPase) and ccca (vacuolar Fe 2+ /Mn 2+ transporter) were all found to be significantly upregulated in the MMrMsn2 mutants (Fig. 5B).
MrMsn2 is required for the virulence of M. rileyi Pathogenicity assays were conducted using third-instar Spodoptera litura larvae. These assays showed that the virulence of the MMrMsn2 mutants was significantly lower than of the WT and CP strains (Fig. 6). The mean lethal time (LT 50 ) for the WT strain was 6.2 AE 0.4 days A. Quantitative analysis of dimorphic transition (from yeast cells to hypha) rate of wild-type, complemented and MMrMsn2 mutants with approximately 100 single yeast cells plated on SMAY medium. The growth morphology was observed every day. B. Median transition time required for 50% transition of blastospores to hyphae (TT 50 ) of wild-type, complemented and MMrMsn2 mutants was compared. TT 50 was calculated using a probit analysis with the SPSS program. Error bars are standard error. * P < 0.05, ** P < 0.01, significantly different compared with wild type. in a topical bioassay and 4.3 AE 0.5 days in the injection bioassays. The LT 50 values for the CP strains were 6.3 AE 0.5 days in topical bioassays and 4.9 AE 0.5 days in injection bioassays, whereas the LT 50 values for the MMrMsn2 mutants were 10.1 AE 0.5 (P < 0.001) in the topical bioassay and 8.6 AE 0.4 (P < 0.001) in the injection bioassays. These results show that MrMsn2 is required for virulence.

Discussion
In our previous investigation, we found that the HOG signalling pathway regulated the dimorphic transition and MS development (Song et al., 2016b). To better define the mechanisms of regulation, in this study, we identified and characterized the transcription factor MrMsn2, which is predicted to be downstream of the HOG pathway in Fig. 3. Pigment produced by wild-type, complemented and MMrMsn2 mutant strains and RT-qPCR of pigment biosynthesis, chitin synthase and antioxidant enzyme genes. A. Bottom of morphology clone of wild-type, omplemented and MMrMsn2 mutants. Bar = 5 mm. B. Quantitative analysis pigment produced. Pigment was calculated from three independent experiments and measured spectrophotometrically by absorbance at 459 nm. RT-qPCR of pigment biosynthesis-related genes (C) and (D) antioxidation genes and chitin synthase genes during conidia development of wild-type or MMrMsn2 mutants. At 3, 6, 9 and 12 days, clones of wild-type or MMrMsn2 mutants were collected for RT-qPCR analysis. Mrtub and Mrtef genes were used as reference. Error bars are standard error. * P < 0.05, ** P < 0.01, significantly different compared with the results at day 3. M. rileyi. The results were unexpected because the gene was important for conidiation, the yeast-to-hypha transition and MS formation.
MrMsn2 belongs to a group of proteins containing C 2 H 2 -like Zn finger domains that are important in development, secondary metabolism and stress responses  (Chang et al., 2011;Liu et al., 2013;Zhang et al., 2014;Tian et al., 2017). In this study, MMrMsn2 mutants negatively controlled the yeast-to-hypha transition (Figs 1 and  2). This result was unlike the observation in Yarrowia lipolytica yeast, in which disruption of Mhy1p, an Msn2/ 4-like protein, restricted the dimorphic transition, and in C. albicans where it had no significant role in CaMsn2/ CaMsn4 mutations (Hurtado and Rachubinski, 1999;Nicholls et al., 2004). These studies show distinct strategies for regulating the yeast-to-hypha transition using Msn2 in different fungi.
Moreover, similar to A. parasiticus and A. flavus (Chang et al., 2011), deletion of MrMsn2 inhibited growth but increased production of conidia (Fig. 1). This was unlike in MMoMsn2 in M. oryzae, MBbmsn2 in B. bassiana or MMrmsn2 in M. robertsii, which decreased conidia production Zhang et al., 2014). Msn2-like proteins do not appear to be important in transcriptional regulation of the stress response in C. albicans and V. dahliae (Nicholls et al., 2004;Tian et al., 2017), but are important in the stress response in C. glabrata, and S. cerevisiae (Schmitt and Mcentee, 1996;Roetzer et al., 2008). Our research found that MMrMsn2 mutants were defective in response to osmotic, cell wall perturbation and oxidative stress (Fig. S3). These studies show distinct strategies for regulating the stress response and conidiation by Msn2 in different fungi.
In eukaryotes microorganism, cellular developmental processes are reported to correlate with increased reactive oxygen species (ROS) levels (Georgiou et al., 2006;Takemoto et al., 2007). Fungi have evolved effective antioxidant mechanisms that include enzyme families that act as ROS scavengers. Previous studies have shown members of antioxidant enzymes families such as SODs and CATs may have complementary effects during the cellular developmental processes (Xie et al., 2012;Youseff et al., 2012;Wang et al., 2013;Zhang and Feng, 2018). Our investigations confirmed that antioxidant enzyme genes cat1, sod1, cat2 and sod2 had different expressions during conidia development in M. rileyi (Fig. 3). In response to ROS stress, Msn2/4  Yi and Huh, 2015). However, deletion of MrMsn2 increased production of ROS (data not shown) and upregulated expression of antioxidant genes to protect against ROS. Upregulation of these antioxidant genes in MMrMsn2 mutants suggests that there may be regulated by other signalling network (Zhang and Feng, 2018). Secondary metabolism, such as pigments production, is triggered and intensified by ROS build-up, with pigments being important for protecting the fungi against stress conditions (Cho et al., 2012;Hong et al., 2013). In the absence of MrMsn2 aggravated pigmentation and pigment synthesis-associated genes were significantly upregulated during conidiation in MMrMsn2 mutants. Chitin is a main component of cell walls and is related to morphogenesis and adaptation to ecological niches (Roncero, 2002;Liu et al., 2017). Expression analysis showed that genes from the class I and II of chitin synthases were downregulated during morphogenesis and conidiation, and then in the absence of MrMsn2, these were upregulated (Fig. 3). Based on this data, we propose a link between chitin biosynthesis and MrMsn2, however, the molecular mechanism remains unknown.
Current conidia mass production methods are not cost-effective, limiting M. rileyi commercialization. MS can be used as an alternative fungal propagule for mycoinsecticide  and has been used in large-scale production in submerged fermentation (Song et al., 2017). As for solid culture, MMrMsn2 mutants were defective in hyphal growth in liquid AM (data not shown). Furthermore, vegetative hyphae are the prerequisite for MS formation (Song et al., 2013;Jiang et al., 2014). Consistent with the defective MS formation in the MMrHog1 mutants, the MMrMsn2 mutants had limited ability to form MS (Fig. 4). This result was unlike the observation in V. dahliae, in which VdMsn2 deletion mutants produce more MS than wild type on solid media (Tian et al., 2017). Previously, it was found that the CWI and HOG signalling pathways cooperate to regulate MS development (Song et al., 2016b). This has been confirmed in this study, with the core genes of the CWI and HOG signalling pathway both being upregulated in the MMrMsn2 mutants. Topical infections were generated by immersing of conidial suspensions of tested strains (5 ll of 1 9 10 7 conidia ml À1 in cotton seed oil). For injection assays, insects were injected conidial suspensions of tested strains (5 ll of 1 9 10 6 conidia ml À1 in sterile water with 0.01% Tween 80). Three replicate groups had 30 larvae each. Controls were treated with pure cotton seed oil or sterile water containing 0.01% Tween 80 only. Error bars are standard error of three trials. * P < 0.05, ** P < 0.01, significantly different compared with wild type. Our previous study demonstrated that intracellular H 2 O 2 levels fluctuated during MS development and peak at MS initiation stage (Song et al., 2018). Antioxidant enzyme genes were significantly upregulated in MMrMsn2 mutants (Fig. 5), indicating that intracellular H 2 O 2 levels were not equilibrated in the MS-initiating formation. This result confirmed our previous results that oxidative stress triggered MS formation (Song et al., 2013(Song et al., , 2016b(Song et al., , 2018Jiang et al., 2014). As mentioned, the signalling network that regulates the chitin synthesis-and pigment synthesis-associated genes in solid SMAY and liquid AM culture is highly complicated. Major basal salts such as iron and calcium cations are necessary for MS formation  and SidA is the major pathway of cellular iron uptake for MS formation (Li et al., 2016). We demonstrated that the iron importer and calcium transports, Ct-1 and Ct-2, were important in MS formation (Wang and Yin, unpublished data). In addition to similar transcriptional mechanisms by Msn2/4 for regulating ccc1 in yeast (Li et al., 2017), we found multiple transcriptional mechanisms for regulating genes for iron and calcium cation transport and storage by MrMsn2 during MS development (Fig. 5). However, the transcriptional mechanisms for regulating cation transport are not clear and further experiments are needed to elucidate the multiple mechanisms.
Research on the function of Msn-like transcriptional factors in pathogenicity is widespread for entomopathogenic, human pathogenic and phytopathogenic fungi (Roetzer et al., 2008;Liu et al., 2013;Zhang et al., 2014). In M. rileyi, our data indicated that MMrMsn2 mutants were significantly less pathogenic than WT by both topical infection and injection assays (Fig. 6). Similar results were reported for MMoMsn2 mutants of M. oryzae, MBbmsn2 mutants of B. bassiana, and MMrmsn2 mutants of M. robertsii in which gene deletions cause decreased virulence Zhang et al., 2014). In contrast, in C. glabrata, Msn2 was found to have no effect on virulence (Roetzer et al., 2008). One explanation for this is that MMrMsn2 mutants counter oxidative stress from hosts in vivo (Song et al., 2016a,b) and were hypersensitive to stress. Another explanation involves morphogenic defects in the mutants. This investigation revealed that vegetative growth of MMrMsn2 mutants was defective in haemocoel (data not shown). These results could be reasons why the mutants had decreased pathogenicity in vivo.
In summary, this study revealed the MrMsn2 had negative effects on the dimorphic transition and conidiation and was required for abiotic stress resistance, virulence, and MS formation. Furthermore, the current transcriptional networks of MrMsn2 during conidia and MS development will enhance our ability to comprehensively understand the molecular mechanism of yeast-to-hypha transition and conidia and MS development.

Gene cloning and bioinformatics analyses
To study the potential function of MrMsn2, the full genomic DNA sequence of MrMsn2 was amplified using primers MrMsn2-L/MrMsn2-R (Table S1) based on sequences of a previous transcriptomic library (Song et al., 2013). The protein sequence and potentially homologues from other fungal species were aligned with Blastp (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Multiple sequences were aligned with DNAMAN software (http:// www.lynnon.com). Neighbour-joining tree was generated using the software MEGA 6.0 (http://www.megasofware. net) (Tamura et al., 2013).

Generation of deletion and complementation mutants
The M. rileyi genome was not annotated when we constructed the targeted gene deletion plasmid. Therefore, fusion primer and nested integrated PCR (Wang et al., 2011) with primers in Table S1 were used to obtain flanking regions (data not shown). Upstream and downstream flanking sequences were amplified using primers Ms-LF/Ms-LR and Ms-RF/Ms-RR (Table S1) M. rileyi genome (Shang et al., 2016) using the primers Ms-HF/Ms-HR (Table S1). PCR products were digested by restriction endonucleases and ligated into the sulfonylurea resistance vector pPZP-Sur-Knock to generate the plasmid, pPZP-Sur-msn2. Disruption and complementation vectors were transformed into Agrobacterium and transformants were screened as described previously (Song et al., 2018).

Phenotypic analyses of test strains on SMAY media
To analyse the function of MrMsn2 in yeast-to-hyphae transition, vegetative growth, conidial development and abiotic stress tolerance, conidial suspensions of the tested strains were plated on SMAY as previously described (Song et al., 2016b). Colony morphology was investigated, and images were collected using a digital camera (60-mm Macro lens; Canon Inc., Tokyo, Japan) and microscope. Conidia numbers were counted as previously described . In vitro, M. rileyi was grown in yeast cell form for 2-4 days on SMAY to transformed into the filamentous form.
The switching rates of the tested strains were counted as previously described (Li et al., 2016). Approximately 100 simple yeast cells were pipetted onto SMAY medium and grown at 25°C. Switching rates at indicated time points were recorded and TT 50 was estimated.

Purification and measurement of clone pigments
Clone pigments of tested strains were counted as described previously (Li et al., 2016). Clones from 6-dayold, 9-day-old or 12-day-old SMAY cultures were isolated with 2% NaOH. Collected clones were ground with liquid nitrogen, suspended in NaOH solution and boiled at 100°C for 2 h. Subsequently, solutions were acidified to pH 2.0 with HCl. Precipitates after centrifugation at 6000 9 g for 15 min, were dissolved in 2% NaOH and measured on a spectrophotometer at the wavelength of 459 nm (Babitskaya et al., 2000).

MS formation assay in liquid AM media
Conidial suspensions of tested strains were inoculated in liquid AM culture for 6 days. Biomass and MS yield were quantified for the AM cultures and determined as previously described . MS morphologies were observed using the digital camera.

Gene expression by qRT-PCR
To assess the effect of exogenous agents on MrMsn2 expression, AM or MM was supplemented with exogenous 1 M iron or 3 mM H 2 O 2 as previously described (Song et al., 2016b(Song et al., , 2018. Mycelia were subsequently harvested for RNA extraction. For time-specific expression patterns during conidia development, samples of WT inoculated on SMAY were collected at 0, 2, 4, 6 and 8 days for RNA extraction. To explore the impact on other genes related to dimorphic transition and melanin production during conidiation, WT or MMrMsn2 mutants were incubated on SMAY media and 3-, 6-, 9-or 12-day-old clones were collected independently for transcriptional analysis. For time-specific expression patterns during MS development, samples of WT inoculated in AM were collected at 36, 60, 72, 96 and 120 h for RNA extraction. To investigate the regulation of other genes during MS formation, WT and MMrMsn2 mutants were incubated in AM cultures. After 72 h, mycelia were collected and total RNA was prepared. Gene expression patterns were confirmed for samples of WT, MMrMsn2 or CP mycelia cultured in AM for 72 h. Total RNA was collected according to previous methods . RT-qPCR was performed using SYBR Green (Invitrogen, Shanghai, China), as per the manufacturer's instructions. b-tubulin (Mrtub) and translation elongation factor (Mrtef) genes were used as internal standards. Relative expression levels were evaluated using the 2 ÀDDCt method (Vandesompele et al., 2002).

Insect virulence assays
Topical infections tests and injection assays were conducted as previously described (Song et al., 2016b). Three replicate groups had 30 larvae each, and after treatment, the larvae were reared as described previously . Larval mortality was recorded daily, and LT 50 values were calculated using probit analysis with the software SPSS 17.0 (SPSS Inc., Chicago, IL, USA).

Data analysis
All assays were repeated three times. Data were analysed by one-way analysis of variance (ANOVA), followed by Duncan's multiple range tests using SPSS 17.0 software. Graphs were constructed in GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA, USA). Error bars represent the standard error.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article.