Unveiling the function and regulation control of the DUF3129 family proteins in fungal infection of hosts

Abstract

Many prokaryotic and eukaryotic proteins contain domains of unknown function (DUFs). A DUF3129 family of proteins is widely encoded in the genomes of fungal pathogens. A few studies in plant and insect pathogens indicated that the DUF3129 genes are required for fungal penetration of host cuticles with an unclear mechanism. We found that a varied number of DUF3129 proteins is present in different fungal species and the proteins are evolutionarily diverged from each other at the inter- and intra-specific levels. By using the insect pathogenic fungus Metarhizium robertsii as a model, we performed experiments and found that the seven DUF3129 proteins encoded by this fungus are localized to cellular lipid droplets (LDs). Individual deletion of these genes did not affect fungal formation of the infection structure appressoria and the accumulation of LDs in fungal conidia. When compared with the wild-type (WT) strain, insect bioassays revealed that the virulence of most null mutants were significantly impaired during topical infection but not during injection of insects. Carbon starvation and the subsequent Western blot analysis indicated that the LD-specific perilipin protein was completely degraded in the WT cells whereas varied levels of perilipin could be detected in the mutant cells, which signified that depletion of LD content was delayed in mutant cells, and DUF3129 proteins are therefore involved in LD degradation. We also provided biochemical evidence that these DUF3129 genes are transcriptionally regulated by a yeast Ste12-like transcription factor. The findings of this study not only unveil the function of DUF3129 proteins but also better understand the diverse mechanism of fungus–host interactions.

This article is part of the theme issue ‘Biotic signalling sheds light on smart pest management’.

1. Introduction

The number of proteins containing domains of unknown function (DUFs) reaches over 4000 families [1], and approximately 40% of the proteins encoded in eukaryotic genomes are DUFs (also called PUFs or POFs, proteins of unknown function) [2]. It has been suggested that common DUFs might play essential roles in organismal physiologies and developments [3]. It is still technically challenging to study the function of these DUFs. Our comparative genomic analysis of plant, insect and mammalian fungal pathogens identified a group of proteins containing the DUF3129 domain (pfam11327) [4]. A literature review indicated that a DUF3129-containing protein gene Egh16 was first cloned from the powdery mildew fungus Erysiphe graminis f. sp. hordei (now named Blumeria graminis f. sp. hordei) [5]. It was later found that multiple copies of Egh16-like genes (at least 10) are present in B. graminis by encoding proteins with variable C-terminus regions, and the proteins were implicated in mediating fungus–plant interactions [6]. Indeed, deletion of the orthologous genes of Egh16 (i.e. Gas1 and Gas2; corresponding to the genes termed Mas3 and Mas1 for Magnaporthe appressoria specific) in the rice blast fungus Magnaporthe oryzae impaired fungal abilities to penetrate host cuticles [7]. Likewise, deletion of a Gas1-like gene Magas1 in the locust-specific pathogen Metarhizium acridum also impaired fungal virulence during topical infection but not during injection of locusts, i.e. the requirement of this gene for insect cuticle penetration [8]. Until this study, the exact function(s) of these DUF3129 proteins has remained unknown.

Ascomycete insect pathogens are evolutionarily closer to plant pathogens than to mammalian pathogenic fungi [4]. The infection of insect hosts by the species such as Metarhizium spp. also requires the formation of appressoria and the build-up of turgor pressor within appressoria to penetrate the protein- and chitin-rich insect cuticles [9]; during which process, the genes required for fungal penetration and therefore the virulence include those involved in mediating lipid droplet (LD) storage/accumulation [10,11], LD biogenesis [12–14], and LD degradation to release glycerol from triacylglycerols [12,13]. Our recent LD proteome study of Metarhizium robertsii identified a DUF3129 protein (MAA_02539) that was yielded in association with cellular LD content [15]. It remains to be determined whether the DUF3129 proteins are localized at the LD surface and involved in mediating LD biogenesis or degradation in fungal cells.

It was found that the appressorium-specific expressions of Gas1 and Gas2 were inactivated in a null mutant of mitogen-activated protein kinase (MAPK) gene Pmk1 [7]. The Pmk1 of Ma. oryzae is a functional orthologue of yeast Fus3, and the transcription factor (TF) Ste12 is one of the substrates phosphorylated by Fus3 [16]. Functional study of a Ste12-like TF (Mst12) in Ma. oryzae indicated that this TF regulates fungal invasive growth and functions downstream of Pmk1 [17]. Deletion of the Ste12-like TF MaSte12 and the subsequent digital gene expression analysis indicated that an array of genes was downregulated in the null mutant of Me. acridum, including the Gas1-like genes [18]. The MAPK pathways have been well investigated in control of conidiation, pathogenicity and stress responses in Me. robertsii [19–21]. It is highly likely that the DUF3129 genes are the downstream effectors of MAPK pathway(s) and Ste12-like TF. However, the biochemical evidence is still missing.

In this study, we report that a varied number of DUF3129 proteins is encoded by different fungal pathogens. By using the insect pathogen Me. robertsii as a model, we found that the seven Gas1-like proteins (termed Mras1-Mras7, for Metarhizium robertsii appressoria specific) encoded by the fungus are localized at the LD surface, implicated in mediating LD degradation and required for fungal penetration of insect cuticles. We also found that DUF3129 family genes are controlled by the Ste12-like TF (termed Mrst12) in Me. robertsii.

2. Material and methods

(a). Fungal strains and cultural conditions

The wild-type (WT) strain and mutants of Me. robertsii ARSEF 2575 were routinely cultured on potato dextrose agar (PDA, Difco) at 25°C. For liquid incubation, fungal conidia were grown in Sabouraud dextrose broth (SDB, Difco) at 25°C on a rotatory shaker. For carbon starvation, fungal cells were grown in the minimum medium without glucose (MM-C) [13]. Artificial induction of appressorium formation was conducted on plastic Petri dishes containing MM plus 1% glycerol (MM+Gly) [11]. Strain EGY48 (Invitrogen) was used for yeast one-hybrid tests. Yeast cells were cultured in YPDA (1% yeast extract, 2% peptone, 2% glucose and 2 mg l⁻¹ adenine), YPD (1% yeast extract, 2% peptone and 2% glucose) or synthetic dropout (SD) agars [14,22]. The BL21 strain of Escherichia coli was used for protein expression.

(b). Phylogeny analysis

To investigate the distribution of DUF3129 proteins encoded in fungal genomes, a genome-wide survey was performed using the sequenced genome information of seven Metarhizium species [23], the selected insect pathogens including Beauveria bassiana, Cordyceps militaris and Ophiocordyceps sinensis [24], and the plant pathogens Ma. oryzae, Fusarium graminearum and B. graminis [25]. The retrieved protein sequences were aligned using Clustal X and a neighbour-joining (NJ) phylogenetic tree was generated using MEGA7 [26] with a Dayhoff model, pairwise deletion for missing data and 1000 bootstrap replications. The DUF3129 domain sequence was also retrieved from each protein for NJ phylogenetic analysis.

(c). Gene deletion and protein localization

Targeted gene deletion of Mras genes was performed by homologous recombination. In brief, the 5′- and 3′-flanking regions of an individual Mras gene were amplified with different primer pairs (electronic supplementary material, table S1). The products were purified and digested with restriction enzymes and then inserted into the corresponding restriction sites of the binary vector pDHt-bar for Agrobacterium-mediated fungal transformation of the WT strain of Me. robertsii [14,27]. To verify whether the Mras genes are controlled by the Ste12-like TF in Me. robertsii, the Mrst12 gene (MAA_05929) was also deleted in this study. To complement the deletion of Mrst12, full length Mrst12 genes were amplified by polymerase chain reaction (PCR) and cloned into the binary vector pDHt-ben for transformation of ΔMrst12 to generate the rescued mutant Comp. Colony PCR and or reverse transcription (RT)-PCR was performed to verify successful gene deletions. Quantitative real-time RT-PCR (qRT-PCR) analysis was performed to determine the expression of Mrst12 when the fungus was grown in different media and the comparative transcription of Mras genes after deletion of Mrst12.

To determine the subcellular localization of Mras proteins, the GFP gene was fused in frame to the 3′-terminus of an individual Mras gene and the cassette was made under the control of the constitutive Hsp70 (MAA_02081) promoter [27]. The products were individually cloned into the binary vector pDHt-bar for transformation of the WT strain. Cellular LD staining was conducted using the LD-specific fluorescent dye of either Nile red (NR) or Bodipy (Thermo Fisher) [10]. For subcellular localization of Mrst12, the Hsp70 gene promoter, GFP and Mrst12 genes were individually amplified with different primer pairs (electronic supplementary material, table S1) and the products were fused by PCR and the cassette was cloned into the binary vector pDHt-bar for transformation of the WT strain of Me. robertsii. Fungal nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI), Sigma).

(d). Appressorium induction and insect bioassays

To determine the effect of the Mras gene on appressorium formation, the conidia of the WT and mutants were either inoculated on the MM+Gly plate or the hind wings of adult mealworms (Tenebrio molitor) for 18 h for microscopic examination [11]. To examine the virulence contribution of Mras genes, insect bioassays were conducted by topical infection and injection of the last instar larvae of the wax moth Galleria mellonella. Conidia of the WT and mutant strains were harvested from the two-week old PDA plates and applied topically by immersion of insects for 30 s in an aqueous suspension of 1 × 10⁷ conidia ml⁻¹. For the injection assay to bypass insect cuticles, an individual insect was injected from the second proleg with 10 µl of spore suspension (1 × 10⁶ conidia ml⁻¹). Both topical and injection bioassays were also conducted to compare the virulence difference between WT and Mrst12 mutants using the 5th instar larvae of silkworm Bombyx mori. Each treatment had three replicates with 15 insects each and the experiments were repeated twice. Control insects were treated with 0.05% Tween-20 and mortality was recorded every 12 h. The median lethal time (LT₅₀) was calculated and compared using Kaplan–Meier analysis with the program SPSS (v. 21.0) [12].

(e). Western blotting

The turnover of the LD-specific perilipin protein Mpl1 is associated with the LD content in fungal cells [10,15]. To reflect cellular LD content after deletion of Mras genes, Western blot analysis was performed by using the antigenic peptide (RADSLGDKTLDRIDERFPIVKKPTS) antibody raised before to detect perilipin Mpl1 [14,28]. The conidia of the WT and mutant were germinated in SDB for 36 h, and the samples were harvested and washed with sterile water three times. An aliquot of each sample was further grown in MM-C for 24 h for carbon starvation to trigger LD degradation [15]. Total proteins were extracted from both SDB and MM-C samples with the cell lysis buffer (Beyotime, China). Aliquots (10 µg each) of proteins were loaded for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel analysis. Western blotting was performed as we described before [14].

(f). Transcriptional control analysis

To further determine the transcription control of Mras genes by Mrst12 in Me. robertsii, both the yeast one-hybrid test and electrophoretic mobility shift assay (EMSA) were performed. Thus, the full-length cDNA of Mrst12 was cloned into the yeast vector pPC86 and the promoter regions of Mras genes were amplified and cloned into the vector p178 for co-transformation of the yeast strain EGY48. The Trp prototrophic transformants were selected using a SD-Trp medium, and the positive colonies were transferred into the SD-Trp-His-Ade plates for transcriptional activation assay [27]. A negative control was made by transforming the yeast cells with a blank vector while positive-control yeast was transformed with the pGBKT7 vector containing a Gal4 activation domain sequence [22]. For EMSA analysis, Mrst12 protein was first expressed in E. coli using the expression vector pET28a and purified using a Ni-NTA agarose (Qiagen, USA). The promoters of Mras genes were amplified using the Cy5-labelled primers (electronic supplementary material, table S1). The unlabelled primers were used for amplification of cold probes for competition assays [29]. The consensus binding motif of Ste12 is YGAMAMR [30]. The mutated motifs CGACAAG and AAAAAA were generated by PCR with primers (electronic supplementary material, table S1) and used for additional control analysis. DNA gels were scanned with the Starion FLA-9000 system (FujiFilm, Japan).

3. Results

(a). Divergent evolution of DUF3129 proteins

Current curation of DUF3129 (Pfam11327; IPR021596) family proteins indicates that they are only present in the genomes of ascomycete and basidiomycete fungi (http://pfam.xfam.org/family/PF11327). However, the number of DUF3129 proteins varies substantially among fungal species. Our genome survey identified seven proteins from Me. robertsii each containing a DUF3129 domain. These proteins are termed Mras1-Mras7: Mras1 for MAA_02539, Mras2 (MAA_08289), Mras3 (MAA_07052), Mras4 (MAA_02880), Mras5 (MAA_05111), Mras6 (MAA_07502), and Mras7 (MAA_05223). Both the full length and DUF3129 domain sizes vary between proteins (figure 1a). Besides seven DUF3129 proteins encoded in Me. robertsii and other Metarhizium species, only four are present in Metarhizium album and six in Metarhizium brunneum [23]. More significant number variations were found among the plant pathogenic fungi, e.g. four in F. graminearum, seven in Ma. oryzae whereas 19 in B. graminis (figure 1b). It is noteworthy that none or only one DUF3129 protein is present in the genomes of saprophytic fungi [25]. Phylogenetic analysis of Mras1-Mras7 together with those DUF3129 proteins retrieved from the selected insect and plant pathogens indicated that, at least, eight linages could be evident. Interestingly, clear divergence of these proteins was observed between and within the insect and plant pathogens (figure 1b). For example, Mras1-Mras3, Mras5 and Mras6 are specific to insect pathogens, whereas those from plant pathogens formed species-specific lineages (or sub-lineages). We also performed phylogeny analysis using the DUF3129 domain sequences and the obtained tree is lineage-specifically congruent with the full-length protein sequence tree (figure 1b and electronic supplementary material, figure S1). The results suggested that the evolution trajectory of these proteins was largely directed by the DUF3129 domain.

Figure 1.

Characterization and phylogenetic analysis of DUF3129 family proteins. (a) Schematic structuring of the DUF3129 proteins encoded in Me. robertsii. (b) Phylogenetic analysis of DUF3129 family proteins from the selected insect and plant fungal pathogens. Protein tags of insect pathogens are: MAC for Metarhizium acridum; MAN, Metarhizium anisopliae; MBR, Metarhizium brunneum; MGU, Metarhizium guizhouense; MAJ, Metarhizium majus; BBA, Beauveria bassiana; CCM, Cordyceps militaris; OCS, Ophiocordyceps sinensis. Protein tags of plant pathogenic fungi (genes labelled in green) are: MGG for Magnaporthe oryzae; FGSG, Fusarium graminearum; CCU, Blumeria graminis f. sp. hordei. (Online version in colour.)

(b). Localization of Mras proteins on lipid droplets

To verify the LD localization of Mras proteins, an individual Mras gene was fused with a GFP gene to transform the WT strain of Me. robertsii. After staining the conidia with the LD-specific dye NR, we found that the GFP signal was overlapped with LDs in all obtained transformants (figure 2). After induction of appressorium differentiation on the hydrophobic surface, NR staining also revealed that the Mras proteins were present on LDs either in the mother conidia or in appressorial cells (electronic supplementary material, figure S2). Thus, consistent with the previous LD proteome data [15], the DUF3129 proteins Mras1-Mras7 are present on the surface of LDs in Me. robertsii.

Figure 2.

Localization of Mras protein on the LD surface in fungal conidia. The conidia of the mutants expressing the Mras-GFP fusion protein were harvested from two-week old PDA plates and stained with the dye Nile red (NR). BR, bright field images. Bar, 5 µm.

(c). Mras genes are required for fungal virulence by contributing to cuticle penetration

To further determine the function of Mras1-Mras7, these genes were individually deleted by homologous replacements. Consistent with the observations in Ma. oryzae [7], no obvious phenotypic differences were observed between WT and mutants when grown on a PDA medium (electronic supplementary material, figure S3). We then performed insect bioassays by both topical infection and injection of the last instar larvae of the wax moth. Statistical comparison of the LT₅₀ values indicated that no obvious difference was observed between WT and null mutants during injection assays. However, except for ΔMras4, a significant difference (p < 0.0001) was evident between WT and other mutants after topical infection (table 1). The data suggested that Mras genes were required, if not all, for fungal virulence by contributing to penetrate insect cuticles.

Table 1.

Estimation and comparison of the LT₅₀ values between the WT and Mras null mutants against the last instar larvae of the wax moth.

strains	topic infection		injection
	LT50 (days)	statistic testa	LT50 (days)	statistic testa
WT	2.23 ± 0.06	—	1.96 ± 0.06	—
ΔMras1	3.64 ± 0.12	χ2 = 64.06; p < 0.0001	2.01 ± 0.07	χ2 = 0.46; p = 0.4954
ΔMras2	2.80 ± 0.11	χ2 = 19.97; p < 0.0001	2.08 ± 0.07	χ2 = 1.99; p = 0.1577
ΔMras3	2.84 ± 0.11	χ2 = 21.01; p < 0.0001	2.10 ± 0.07	χ2 = 2.23; p = 0.1355
ΔMras4	2.39 ± 0.10	χ2 = 2.43; p = 0.1193	2.06 ± 0.08	χ2 = 1.34; p = 0.3476
ΔMras5	2.96 ± 0.13	χ2 = 22.96; p < 0.0001	2.23 ± 0.07	χ2 = 3.62; p = 0.0533
ΔMras6	2.74 ± 0.11	χ2 = 17.82; p < 0.0001	2.02 ± 0.07	χ2 = 0.89; p = 0.3465
ΔMras7	2.77 ± 0.12	χ2 = 16.52; p < 0.0001	2.11 ± 0.10	χ2 = 2.94; p = 0.0863

^aLog-Rank statistic tests were performed between the WT and individual mutant.

(d). Mras genes are not required for appressorium differentiation but lipid droplet degradation

Successful formation of the infection structure appressoria and the build-up of proper appressorial turgor are required for the cuticular penetration and host infection in Metarhizium species [9]. We performed the induction of appressorium formation and found that, relative to the WT, no obvious difference could be observed between the WT and individual mutant both on an artificial medium MM+Gly and on the insect hind wings (electronic supplementary material, figure S4). Staining of fungal conidia with the LD-specific dye Bodipy indicated that the cellular LD contents of Mras gene mutants were similar to those of WT cells (figure 3a), indicating that Mras proteins have no contribution in mediating cellular LD biogenesis and storage. We further performed experiments to examine whether the LD-localized Mras protein would be involved in mediating the degradation of LDs, i.e. the LD-content reduction and conversion of TAG to glycerol for the generation of appressorium turgor [10,12]. Western blot analysis using the Mpl1 antibody indicated that the perilipin protein was similarly detected between the WT and mutant cells after growth in a nutrient-rich SDB medium (figure 3b). After carbon starvation, however, cellular perilipin disappeared in the WT cells, whereas varied levels of Mpl1 could be detected in the cells of ΔMras1, and ΔMras3-ΔMras7 (figure 3c). The results revealed that, relative to the WT, LD content reduction was not obviously observed after deletion of Mras1 and Mras3-7. The observation of the fungal early infection process also demonstrated that the formation of melanization (penetration) spots [31] was impaired in the null mutants when compared to WT. In particular, no penetration spot was found on the insects treated with the conidia of ΔMras3 and ΔMras7 (figure 3d). Taken together, the data suggested that the DUF3129 proteins play essential roles in mediating LD degradation to contribute to fungal virulence.

Figure 3.

LD storage, degradation and implication in fungal penetration of insect hosts. (a) Similar level of LD content observed in the conidia of the WT and mutant strains. The conidia were harvested from two-week old PDA plates and stained with the LD-specific dye Bodipy. Bar, 5 µm. (b) Western blot analysis showing the similar level of perilipin protein Mpl1 present in the mycelial cells of the WT and null mutants harvested from SDB for 36 h. (c) Western blot analysis showing the varied level of Mpl1 protein present in the mycelial cells of the WT and mutants after starvation for 24 h. Parallel SDS-PAGE protein gel analysis was used as a reference showing the loading of 10 µg proteins for each sample. (d) Impairment on host cuticle penetration after deletion of Mras genes. The insects were topically infected for 48 h. Relative to the WT, fewer or none melanization spots (indications of fungal penetration events) were observed on the insects treated with the mutant conidia. Bar, 1 cm.

(e). Mras genes are regulated by Mrst12

To verify the transcriptional control of Mras genes by Ste12-like TF, we first performed a genome survey and identified a single yeast Ste12-like gene in Me. robertsii (MAA_05929; 60% identity at the amino acid level), designated Mrst12. After growing the fungus in different conditions, qRT-PCR analysis demonstrated that Mrst12 was highly activated during fungal formation of appressoria (figure 4a). The fusion experiment of Mrst12 with a GFP protein confirmed that Mrst12 is localized in nuclei (electronic supplementary material, figure S5), the support for Mrst12 to function as a TF. After deletion of Mrst12 and generation of its rescued mutant Comp, both the WT and mutants were induced for appressorium formation. In contrast to the WT and Comp, ΔMrst12 failed to form appressoria on an artificial medium whereas its differentiation on insect wings was not substantially affected (electronic supplementary material, figure S6). Insect bioassays against the silkworm larvae revealed that the null mutant of Mrst12 could not cause 50% mortality during topical infection. However, no obvious difference was observed between WT and mutants during injection assays (electronic supplementary material, table S2).

Figure 4.

Regulation of Mras genes by MrSt12 in Me. robertsii. (a) qRT-PCR analysis of Mrst12 expression. RNA was extracted from the conidia (CO) harvested from PDA for two weeks, mycelia (MY) harvested from SDB for 36 h, appressoria (App) formed on the insect hind wings for 18 h and the hyphal bodies (HB) harvested from the insect haemocoels 48 h post injection. ***p < 0.001. (b) qRT-PCR verification of Mras gene expressions. The WT, ΔMrst12 and Comp were inoculated on the insect wings for 18 h to induce appressorium formation and the RNA was then extracted for gene expression analysis. A β-tubulin gene was used as a reference. The significance level of difference is at: *p < 0.05; ***p < 0.001. (c) Yeast one-hybrid analysis of Mras gene transcriptions controlled by MrSt12. Top panels: yeast clones grown on a SD-Trp medium. Lower panels: yeast clones grown on a SD-Trp-His-Ade medium. PC, positive control; NC, negative control. (d) EMSA analysis of Mrst12 binding to the promoter regions of Mras genes. EMSA was performed by incubating the indicated amounts of purified Mrst12 protein with a Cy5 fluorescence-labelled DNA fragment (1 nM). Competition was achieved by adding 10 nM of unlabelled (cold) probe.

After induction on insect wings, RNA was extracted from the WT and Mrst12 mutants for qRT-PCR analysis of Mras gene expressions. The results revealed that the transcription of Mras1-7 was significantly reduced in ΔMrst12 when compared with their expressions in the WT and Comp strains (figure 4b). In particular, relative to the WT, the expressions of Mras3, Mras4 and Mras6 and Mras7 were down-regulated more than 2-fold (p < 0.001). Consistently, yeast one-hybrid analysis demonstrated that Mrst12 could bind the promoter regions of Mras genes albeit at the variable levels (figure 4c). In particular, more tight binding activities were observed between Mrst12 and the promoters of Mras3 and Mras7 than between Mrst12 and other gene promoters. We also expressed and purified Mrst12 protein (electronic supplementary material, figure S7a) and performed EMSA analysis. It was found again that Mrst12 could bind the Mras gene promoters (figure 4d). To be confirmed, Mrst12 could not bind the mutated motifs (electronic supplementary material, figure S7b). Thus, the DUF3129 genes are regulated by Mrst12 in Me. robertsii.

4. Discussion

Functional study of DUFs is an important endeavour to delineate the complex biologies. However, it remains challenging to do so even in modern biology. In this study, we performed experiments to unveil the function of a family of DUF3129 proteins in the insect pathogenic fungus Me. robertsii. Protein localization assays indicated that all these proteins could target the LDs in fungal cells. Gene deletions revealed that LD content was not affected in the conidia of null mutants when compared with the WT. However, LD degradation and content reduction was impaired to the varied levels in different mutants after carbon starvation. Consistent with this observation, insect bioassays demonstrated that the virulence reduction and impairment on cuticular penetration were evident in the null mutants during topical infection instead of injection assays. It has also been verified that the DUF3129 genes are regulated by the Ste12-like TF in Me. robertsii. The findings in this study greatly facilitate the understanding of the functions of this group of poorly characterized proteins.

Providing DUF3129 proteins are ubiquitously distributed in ascomycete and basidiomycete fungal pathogens, substantial number variations are observed among different fungal species. Number variation of homologous protein families has been frequently observed across different organisms [20]. For example, the number of the chitin-binding LysM effector genes varies even between the closely related insect or plant fungal pathogens [32]. The species of Metarhizium genus evolved from the host-specific species (e.g. Me. acridum and Me. album) to generalists that can infect a broad range of insect hosts (Me. robertsii and Me. anisopliae) along with protein family expansions [23]. Since the varied numbers of DUF3129 is also present between the basal species of Me. album (four) and Me. acridum (seven), it would suggest that the number variation of DUF3129 might not be associated with fungal host specificity in Metarhizium species and other fungal pathogens as well. Considering the requirement of DUF3129 for fungal virulence [7,8], the absence of this protein family in saprophytic fungi would imply the occurrence of gene-loss events during saprophytic fungal speciation. The extensive number increase and formation of a species-specific lineage of DUF3129 in the pathogens like B. graminis would suggest the event of gene duplications. Overall, similar to other protein families, the evolution of DUF3129 gene number variation in different fungal species is still enigmatic.

In Ma. oryzae, the expression of Gas1 and Gas2 genes was found to be specific in appressoria and the encoded proteins were shown to localize in cytoplasm [7]. Consistent with our previous LD proteomic analysis [15], we found that the Gas1- and Gas2-like DUF3129 proteins are localized at the LD surface. It is evident that DUF3129 proteins are required for fungal virulence by their contribution of breaching host cuticles. Successful penetration of insect and plant cuticles requires the formation of infection structure appressoria and generation of proper turgor within appressorial cells [9,33]. Similar to the findings in the rice blast fungus [7] and Me. acridum [8], we found that the DUF3129 proteins Mras1-Mras7 are dispensable for the formation of appressoria. Thus, the mutant failure to breach host cuticles could be owing to the defect of turgor generation. The build-up of appressorium turgor requires the accumulation of compatible solutes, mainly the glycerol [34,35]. The ubiquitous organelle LDs store neutral lipids such as triacylglycerols and steryl esters, and different genes are involved in LD biogenesis and degradation to sequentially produce diacylglycerol, monoacylglycerol and then glycerol [12,36]. Perilipin controls cellular LD size and storage, which can be used as a biomarker to reflect cellular LD content [37]. Carbon starvation could lead to the consumption of LD contents [15]. We found that, except for Mras2, deletion of other Mras genes more or less delayed the proteolysis of perilipin Mpl1 after carbon starvation, i.e. the indication of LD degradation failure. Considering that the LD content in conidia was not affected after deletion of Mras genes, it could be concluded therefore that DUF3129 proteins are involved in mediating LD degradation and subsequently control appressorial turgor. It nevertheless remains to be investigated why different Mras protein play a role of different magnitude and whether there is any functional redundancy of these proteins. It is noteworthy that, similar to the WT, Mpl1 was not detected in ΔMras2 cells after carbon starvation. This observation would suggest that Mras2 might not contribute to LD degradation during fungal saprophytic growth since the virulence of ΔMras2 was also significantly reduced when compared with the WT of Me. robertsii. The molecular mechanism(s) of Mras protein contributions to LD degradation requires further investigation.

Consistent with a previous gene expression analysis after deletion of MaSte12 in Me. acridum [18], our yeast one-hybrid and EMSA analyses verified that Mras1-Mras7 are regulated by the yeast Ste12-like TF Mrst12 in Me. robertsii. However, it was found that the binding and activation activities of Mrst12 varied when targeting the promoter of different Mras genes. The expression of Mrst12 was more highly activated in appressoria than in other type of cells. It cannot be precluded therefore that other TF(s) might jointly control the expression of divergent DUF3129 genes. Yeast Ste12 is the downstream effector of the MAPK Fus3 [16]. It has been shown that the yeast Fus3-like MAPK is required for fungal virulence in Me. robertsii [19]. Future studies are still required to determine whether the Fus3-like MAPK controls Mrst12, Mras1-7 and beyond.

In conclusion, we revealed that DUF3129 proteins are localized to LDs and are involved in LD degradation. Based on this understanding, we would like to define this domain as TLDD for Targeting Lipid Droplets for Degradation. Fungal virulence contribution of these genes can be postulated to their participations in generation of the appressorium turgor for breaching host cuticles. We also provided the biochemical evidence that this family of proteins are regulated by the Ste12-like transcription factor, the downstream substrate of the Pmk1/Fus3 MAPK pathway. The results of this study facilitate the understanding of the diverse mechanisms of fungus–host interactions.

Data accessibility

This article has no additional data.

Authors' contributions

W.H., S.H., G.T. and Y.L. performed experiments; W.H. carried out gene deletions, data analysis and drafted the manuscript; S.H. performed Western blot analysis; G.T. and Y.L. participated in insect bioassays and data collection; C.W. conceived of the study, coordinated the experiments and wrote the manuscript.

Competing interests

The authors have no competing interests.

Funding

This study was supported by the Strategic Priority Research Program of Chinese Academy of Sciences (XDB11030100) and the National Key R&D Program of China (2017YFD0200400).