Glycerol-3-Phosphate Acyltransferase Contributes to Triacylglycerol Biosynthesis, Lipid Droplet Formation, and Host Invasion in Metarhizium robertsii

Abstract

Enzymes involved in the triacylglycerol (TAG) biosynthesis have been well studied in the model organisms of yeasts and animals. Among these, the isoforms of glycerol-3-phosphate acyltransferase (GPAT) redundantly catalyze the first and rate-limiting step in glycerolipid synthesis. Here, we report the functions of mrGAT, a GPAT ortholog, in an insect-pathogenic fungus, Metarhizium robertsii. Unlike in yeasts and animals, a single copy of the mrGAT gene is present in the fungal genome and the gene deletion mutant is viable. Compared to the wild type and the gene-rescued mutant, the ΔmrGAT mutant demonstrated reduced abilities to produce conidia and synthesize TAG, glycerol, and total lipids. More importantly, we found that mrGAT is localized to the endoplasmic reticulum and directly linked to the formation of lipid droplets (LDs) in fungal cells. Insect bioassay results showed that mrGAT is required for full fungal virulence by aiding fungal penetration of host cuticles. Data from this study not only advance our understanding of GPAT functions in fungi but also suggest that filamentous fungi such as M. robertsii can serve as a good model to elucidate the role of the glycerol phosphate pathway in fungal physiology, particularly to determine the mechanistic connection of GPAT to LD formation.

INTRODUCTION

The glycerol phosphate pathway is the major pathway for triglyceride biosynthesis in various organisms ranging from bacteria to animals to provide the crucial energy molecules as well as serve as a repository for biosynthesis of fatty acids and phospholipids (1). Among the various enzymes involved in this pathway, glycerol-3-phosphate acyltransferase (GPAT) is the first enzyme and catalyzes the acylation of glycerol 3-phosphate (G3P) that results in lysophosphatidic acid (LPA), which is the precursor for the biosynthesis of phosphatidic acid, diacylglycerol (DAG), and triacylglycerol (TAG) (2, 3). Different isoforms of GPAT have been characterized in yeast, plants, and animals. For example, two redundant GPATs, i.e., GAT1 (GPT2p) and GAT2 (SCT1p), have been identified in the budding yeast Saccharomyces cerevisiae, three copies each in Drosophila melanogaster and Caenorhabditis elegans, four copies each in humans and mouse, and eight copies in the plant Arabidopsis thaliana (3, 4). While most GPATs acylate the sn-1 position of G3P to produce LPA, the GPAT4 and GPAT6 isoforms in A. thaliana are involved in cutin biosynthesis and predominantly esterify acyl groups at the sn-2 position of G3P to produce sn-2 monoacylglycerol (5). In animals, a peroxisomal dihydroxyacetone-phosphate acyltransferase (DHAPAT) can provide an alternate route for LPA production by acylation of dihydroxyacetone phosphate (DHAP) and the subsequent reduction of 1-acyl-DHAP to LPA (6). Yeast GAT1 can also convert DHAP into 1-acyl DHAP (7, 8), and the latter can be further reduced to LPA by 1-acyldihydroxyacetone-phosphate reductase (AYR1) in yeast (9). A fungus-like DHAPAT is still unknown.

Mammalian GPAT1 and GPAT2 are localized on the outer membrane of mitochondria (MT), while GPAT3 and GPAT4 are targeted to the endoplasmic reticulum (ER) (3, 10). A recent study showed that D. melanogaster and mammalian GPAT4 isoforms could be relocalized from the ER to the surface of nascent lipid droplets (LDs), where they mediate LD growth (4). Similarly, GPAT isoforms ACL-4 and ACL-5 from C. elegans are also localized on the ER membrane. However, ACL-6 is an MT-type GPAT, which is required to control MT fusion and nematode oogenesis (10). Yeast GAT1 and GAT2, also called microsomal GPATs, are ER-type GPATs (11) and contribute to polarized cell growth (12). Deletion of either gene does not affect yeast cell growth; however, the double deletion mutants are not viable (8). The GPAT homolog has not been characterized thus far in filamentous fungi.

The ubiquitous insect-pathogenic fungus Metarhizium robertsii is a biocontrol agent used worldwide to control different insect pests (13, 14). Similar to plant pathogens, insect pathogens such as M. robertsii infect hosts by penetrating host cuticles (15), which is mediated by high concentrations of glycerol within the infection structures appressoria (16, 17). It has been shown that high concentrations of glycerol in the appressoria of the plant pathogen Magnaporthe oryzae are largely lipolytic products that originated from stored TAG (ca. 44% of LD components [18]) rather than from carbohydrate sources by on-site biosynthesis (19). It is possible that the GPAT(s) in fungal pathogens also contributes to fungal virulence. To investigate this, we studied the functional characteristics of a GPAT from the entomopathogenic fungus M. robertsii, designated mrGAT (MAA_02162) (20). Our data indicate the presence of a single copy of mrGAT in the genome of M. robertsii, and the gene deletion mutant, although viable, has significantly impaired TAG accumulation, LD formation, and virulence against insect hosts. We also found a single copy of a human DHAPAT (NP_055051)-like protein gene present in the genome of M. robertsii (MAA_02767; 25% identity with human DHAPAT; designated mrDHAPAT). Deletion of this gene did not result in any obvious phenotypic and physiological changes in the mutants compared with the wild-type (WT) strain.

MATERIALS AND METHODS

Strains and culture conditions.

To collect conidia for the experiments, the WT strain ARSEF2575 of M. robertsii was grown on potato dextrose agar (PDA; Difco) at 25°C for 20 days. Spore germination and appressorium induction assays were conducted using locust (Schistocerca gregaria) hind wings or the minimal medium (MM) (NaNO₃, 6 g liter⁻¹; KCl, 0.52 g liter⁻¹; MgSO₄ · 7H₂O, 0.52 g liter⁻¹; KH₂PO₄, 0.25 g liter⁻¹) amended with 1% glycerol as the sole carbon resource (MMGly) (17). For genomic DNA and RNA extractions, fungal spores were cultured in Sabouraud dextrose broth (SDB; Difco) at 25°C and 200 rpm for 3 days in a rotary shaker.

Phylogenetic analysis.

To determine the phylogeny of mrGAT across fungal lineages, the homologs of mrGAT were retrieved from selected fungal pathogens and saprophytes of ascomycetes, basidiomycetes, microsporidia, chytrids, and zygomycetes with well-annotated genomes using BLASTP searches with a cutoff E value of <1e−100. Sequence alignment was conducted using ClustalX 2.0, and a neighbor-joining tree was generated using MEGA 5.2 (21) with a Dayoff amino acid substitution model, a pairwise deletion for missing residues or gaps, and 1,000 bootstrap replicates. Prediction of mrGAT subcellular localization was performed with the program ProtComp (ver. 9.0, Softberry) and TargetP (ver. 1.1) (22).

Gene deletion and complementation.

For functional studies, mrGAT gene was deleted using an Agrobacterium-mediated transformation method as described in our previous study (23). In brief, the 5′- and 3′-flanking regions of mrGAT were amplified by PCR using the genomic DNA as the template with the primer pairs mrGATUF/mrGATUR and mrGATDF/mrGATDR (see Table S1 in the supplemental material). The amplified products were subsequently cloned into the PstI and SpeI restriction sites of the binary vector pDHt-SK-ben (conferring resistance against benomyl) for fungal transformation (23) to create the ΔmrGAT deletion mutant. To complement gene deletion, the mrGAT gene was amplified together with its promoter and the 3′-untranslated region (3′-UTR) with the primers mrGATCompF and mrGATCompR, and the product was subcloned into the SpeI site of the binary vector pDHt-SK-Bar (conferring resistance against ammonium glufosinate) before fungal transformation to obtain the complemented mutant (Comp). Transformants were verified by PCR and reverse transcription (RT)-PCR analyses using primers mrGATF and mrGATR (see Table S1 in the supplemental material). The β-tubulin gene (MAA_02081) was used as the control and amplified using primers TubF and TubR. Deletion of mrDHAPAT was similarly performed using the primer pairs mrDHAPATUF/mrDHAPATUR, and mrDHAPATDUF/mrDHAPATDR, respectively (see Table S1 in the supplemental material).

Examination of protein localization.

To confirm ER localization of mrGAT, the open reading frame (ORF) of mrGAT gene was amplified by PCR with the primer pairs mrGAT-GFP1F and mrGAT-GFP1R to delete the stop codon and include the promoter region of the gene (1,970 bp upstream from the start codon). The Egfp gene was amplified from the plasmid pEGFP (14) with the primers mrGAT-GFP2F and mrGAT-GFP2R (see Table S1 in the supplemental material). The acquired fragments were purified and jointed together by a fusion PCR (23). The product was digested with the restriction enzymes SpeI and EcoRI and cloned into the same enzyme-treated plasmid pDHt-SK-Bar (conferring resistance against glufosinate) for Agrobacterium-mediated transformation (23). The acquired mrGAT-GFP (green fluorescent protein) strain was cultured in SDB for 3 days, and the mycelia were washed twice with Hanks' balanced salt solution (Gibco) before being stained with the fluorescent dye ER-Tracker Blue-White DPX (catalog number E-12353; Invitrogen) for 30 min. The images were taken with a fluorescence microscope, BX51-33P (Olympus).

Appressorium induction and lipid droplet visualization.

Conidia from the WT, ΔmrGAT, and Comp strains were inoculated into individual polystyrene petri dishes (5.5 cm in diameter) containing 2 ml MMGly at a final concentration of 2 × 10⁵ spores ml⁻¹. After incubation for 18 h, spore germination and appressorium differentiation rates were recorded for >100 spores under a microscope. Appressoria were also induced on locust (S. gregaria) hind wings as described previously (24). To visualize and compare the formation of LDs, conidia, germlings, and appressoria from the WT, ΔmrGAT, and Comp strains were washed twice with phosphate-buffered saline (PBS) and then stained with a fluorescent dye, Bodipy (catalog number D-3922; Invitrogen) for 30 min (17). The accumulation of intracellular LDs was observed under a transmission electron microscope (TEM) as described previously (23). Fungal samples were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) at 4°C for 12 h, rinsed three times in phosphate buffer, and fixed overnight in 1% osmium tetroxide in 0.1 M cacodylate buffer (pH 7.0) at 4°C. After rinsing three times in phosphate buffer, samples were dehydrated in an ethanol gradient series, infiltrated with a graded series of epoxy resin in epoxy propane, and then embedded in Epon resin and sectioned. The ultrathin sections were stained in 2% uranium acetate followed by lead citrate and visualized under a transmission electron microscope (H-7650; Hitachi, Tokyo, Japan) operating at 80 kV.

Free-glycerol and -triacylglycerol assays.

Conidia were collected from 2-week-old cultures on PDA plates, while mycelia were collected from 3-day-old cultures in SDB. Glycerol content in the samples was assayed using the free-glycerol assay kit E1012 (Applygen Technologies Inc., Beijing, China) at 550 nm, while triglycerides were assayed using the triglyceride assay kit E1014 (Applygen Technologies Inc., Beijing, China) at 550 nm. First, all samples were washed twice with phosphate buffer, homogenized in extraction buffer, and centrifuged at 5,000 × g for 5 min. Supernatants were aliquoted and incubated with the reaction buffer for 30 min before determining the optical density at 550 nm (OD₅₅₀) using a microplate reader (Varioskan Flash Multimode Reader; Thermo Scientific). Total protein concentration in supernatants was estimated using the Bradford method. All experiments were repeated twice, and three replicates were maintained for each sample. Glycerol and triglyceride concentrations in different samples were expressed as micromoles of glycerol or triglyceride per milligram of total proteins, respectively.

Total lipid quantification.

To determine the effect of mrGAT on lipid biosynthesis, total lipid was quantified by a phosphoric acid-vanillin method (25). Conidia from WT, ΔmrGAT, and Comp strains were harvested from the PDA plates incubated at 25°C for 20 days, and mycelia were collected from the SDB after incubation at 220 rpm and 25°C for 3 days. For assays, spore suspensions (0.5 ml, 1.0 × 10⁸ conidia ml⁻¹) were added to glass tubes and 1 mg mycelium homogenates (dried overnight at 150°C) was added to glass tubes with 0.5 ml water. Then, 2 ml of 18 M H₂SO₄ was added to each tube, boiled in a water bath for 10 min, and cooled for 5 min at room temperature before adding 5 ml phosphoric acid-vanillin reagent (1.2 g liter⁻¹ vanillin and 200 ml of water, adjusted to 1 liter with 85% H₃PO₄). The tubes were then incubated at 37°C for another 15 min and centrifuged to determine absorbance at 530 nm (17). All experiments were repeated twice. A standard curve was generated using triolein (Sigma) for quantification.

Western blot analysis.

Proteins were extracted from fungal conidia and mycelia with the RIPA lysis buffer (Thermo Scientific) containing 1 mM protease inhibitor phenylmethylsulfonyl fluoride (PMSF). Proteins were separated on 12% sodium dodecyl sulfate containing polyacrylamide gels and transferred onto polyvinylidene difluoride membranes. The membranes were probed with the antibodies against the LD surface perilipin protein, MPL1, and β-tubulin separately (23).

Insect bioassays.

To investigate the effect of mrGAT on fungal virulence, insect bioassays were conducted on second-day, fifth-instar Bombyx mori larvae. Conidia from WT, ΔmrGAT, Comp, and ΔmrDHAPAT strains were applied topically by immersing the larvae in an aqueous suspension containing 5 × 10⁶ conidia ml⁻¹ for 1 min or by injection of a 10-μl suspension containing 1 × 10⁶ spores ml⁻¹ into the second proleg. Each treatment had three replicates with 15 insects each, and all experiments were repeated three times. Larval mortality was recorded every 12 h, and the median lethal time (LT₅₀) was estimated by Kaplan-Meier analysis using SPSS (ver. 13.0).

RESULTS

Characteristics of the mrGAT protein.

The complete ORF of mrGAT encodes a protein with 766 amino acids and a predicted molecular mass of 84.9 kDa and pI of 9.63. MrGAT is a typical member of the Pfam 01553 family of acyltransferases and is homologous to yeast GAT1 (GPT2p, 34% identity) and GAT2 (SCT1p, 37% identity). Similar to other organisms, it contains the four highly conserved AGPAT (1-acyl-sn-glycerol-3-phosphate acyltransferase) motifs for catalysis (motifs I and IV) and G3P binding (motifs II and III) (see Table S2 in the supplemental material). In silico analysis with the programs TargetP and ProtComp indicated with a high confidence score that mrGAT is localized to the endoplasmic reticulum. Survey of the M. robertsii genome revealed the presence of only one copy of the gene, which is similar to what is observed in other fungi from the subphylum Pezizomycotina (phylum Ascomycota) (Fig. 1). A single copy of GPAT was also found in the fission yeast (Schizosaccharomyces pombe) and in a chytrid (Batrachochytrium dendrobatidis). In contrast, two copies were found in the subphylum Saccharomycotina (Ascomycota), which includes the budding yeast, and four copies were found in the zygomycete fungus Rhizopus delemar (Fig. 1). In the basidiomycete fungal species, either one or two copies of the GPAT gene were found. Interestingly, a GPAT gene homolog was absent in the genomes of the microsporidian fungi. Relative to the well-established fungal tree of life (26), the GPAT gene evolutionary tree is not congruent with fungal speciation phylogeny across the phyla. For example, chytrids are in general considered to have diverged early to form a basal clade within the fungal kingdom, while the GPAT tree is rooted by yeast GPAT and the chytrid GPAT is more closely related to that of ascomycetes than zygomycetes (Fig. 1). A genome survey of mammalian DHAPAT protein-like genes showed that a single copy of the gene is present in the genomes of M. robertsii and other ascomycete species of subphylum Pezizomycotina but not in the yeast species of Saccharomycotina (see Fig. S1 in the supplemental material).

Fig 1.

Phylogenetic analysis of fungal glycerol-3-phosphate O-acyltransferases. Protein sequences were retrieved and aligned to generate a neighbor-joining tree with a Dayoff amino acid substitution model. Values above the branches were estimated based on 1,000 bootstrap replicates.

Characterization of the mutants.

In S. cerevisiae, GAT1 and GAT2 are functionally redundant because viability is not affected by deletion of either gene; however, double deletions are lethal (8). To determine the effect of mrGAT on fungal viability, gene deletion and complementation were performed via Agrobacterium-mediated transformation, and the resultant strains were analyzed by PCR and RT-PCR (Fig. 2A and B). We found that the ΔmrGAT strain of M. robertsii was viable, although the ability to sporulate and form pigments was impaired compared to the WT and complemented (Comp) strains (Fig. 2C and D). For example, after growth on PDA for 14 days, the sporulation of the ΔmrGAT mutant (1.47 × 10⁶ ± 0.25 × 10⁶ conidia cm⁻²) decreased drastically (P < 0.01) compared to the WT (22 × 10⁶ ± 5.046 × 10⁶ conidia cm⁻²) and Comp (25.64 × 10⁶ ± 1.45 × 10⁶ conidia cm⁻²) strains. This effect also persisted in the mutant after growth for up to 20 days (Fig. 2D). Consistent with the above in silico analysis, mrGAT was confirmed to target to the ER by green fluorescent protein (GFP) fusion and ER-specific staining (see Fig. S2 in the supplemental material).

Fig 2.

Gene disruption, complementation, and phenotyping in Metarhizium robertsii. (A) PCR confirmation. Genomic DNAs extracted from the wild-type (WT), ΔmrGAT, and gene complemented mutant (Comp) strains were used as the templates for PCR. CK, negative control using water as the template. (B) RT-PCR verification of mrGAT gene in WT, ΔmrGAT, and Comp strains. Tub, β-tubulin gene. (C) Phenotypic characterization. In contrast to WT and Comp, the ΔmrGAT mutant had impaired conidia (upper panels) and pigment production (lower panels show the reverse sides of the plates) after growth on PDA for 2 weeks. (D) Quantification of conidial production by WT, ΔmrGAT, and Comp strains after growth on PDA for 14 or 20 days.

Deletion of mrDHAPAT gene in M. robertsii did not result in obvious phenotype changes, including nonsignificant differences of conidiation and insect-killing ability against the silkworm larvae between the WT and mutant (see Fig. S3 in the supplemental material). In addition, we did not find obvious alternation in LD formation in the mutant cells compared with the WT (see Fig. S4 in the supplemental material). For total lipid content, a difference was found between the WT and ΔmrDHAPAT conidial samples (P = 0.0127) but not between their mycelial samples (P = 0.0948) (see Fig. S5 in the supplemental material). It is noteworthy that repeated trials failed to obtain the mrGAT and mrDHAPAT double deletion mutants in M. robertsii, implying a lethal effect.

Effects of mrGAT on the biosynthesis of triacylglycerol, glycerol, and total lipid content.

The first step in the TAG biosynthesis pathway is the catalysis of G3P to LPA by GPAT (1, 4). Not surprisingly, cellular accumulation of TAG was significantly (P < 0.001) reduced in the conidia and mycelia of the null mutant compared to WT and Comp (Fig. 3A). In addition, cellular glycerol in the ΔmrGAT mutant was also significantly (P < 0.01) lower than in WT and Comp (Fig. 3B). Quantification of total lipids in the conidial samples showed that the ΔmrGAT mutant had only 52% (19.51 ± 0.22 μg 10⁻⁷ spores) of the total lipid content in the WT (37.25 ± 0.18 μg 10⁻⁷ spores) (P = 7.71−6e) and 67% of that in Comp (28.87 ± 0.38 μg10⁻⁷ spores) (P = 2.64−5e), respectively (Fig. 3C). In mycelia, the total lipid content of the ΔmrGAT mutant (40.00 ± 0.85 μg mg⁻¹, dry weight) was also significantly lower than that of WT (59.38 ± 1.01 μg mg⁻¹, dry weight; P = 0.0011) and that of Comp (53.02 ± 0.91 μg mg⁻¹, dry weight; P = 0.0052) (Fig. 3D).

Fig 3.

Quantification of triacylglycerol (TAG), glycerol, and total lipids. (A) Conidia harvested from PDA after 20 days of culture and mycelia collected from SDB after 3 days of culture were used for TAG analysis to demonstrate the differences among the WT, ΔmrGAT, and Comp strains. (B) Differences in glycerol content among the WT, ΔmrGAT, and Comp strains. (C) Differences in total lipid content in conidia among the WT, ΔmrGAT, and Comp strains. (D) Total lipid content variations in mycelia among the WT, ΔmrGAT, and Comp strains. MDW, mycelium dry weight.

Effect of mrGAT on lipid droplets formation.

In eukaryotic cells, lipids such as neutral triglycerides are stored in LDs (27). To examine and compare the formation of LD in the WT and mutant strains, TEM and fluorescent staining assays were conducted. The results indicated that, compared to WT and Comp, the ΔmrGAT mutant stored far fewer LDs in the conidia (Fig. 4A to C; Fig. 5A) and no visible LDs were observed in ΔmrGAT mycelia (Fig. 4D to F; Fig. 5C). Relative to WT, fewer LDs were also observed in the mutant germlings (Fig. 5B) and appressoria (Fig. 5D and E). Formation and stabilization of LDs are essentially controlled by LD surface perilipin proteins, such as the Mpl1 protein in M. robertsii (17). Since the LD formation was impaired in ΔmrGAT, we compared the accumulation pattern of Mpl1 between the WT and null mutant. Western blot analysis demonstrated the highly reduced accumulation of Mpl1 in the conidia and mycelia of the ΔmrGAT strain compared to the WT and Comp (Fig. 4G), which is consistent with the failure in LD formation. This indicated an association between mrGAT and the perilipin protein MPL1.

Fig 4.

Visualization of cellular lipid droplets (LDs) and Western blot analysis of LD surface protein. Conidia from the WT, ΔmrGAT, and Comp strains harvested from PDA plates after growth for 20 days were used for TEM analysis. In contrast to what was observed with the WT (A) and Comp (C) strains, accumulation of LDs (black arrows) was significantly reduced in the ΔmrGAT mutant (B). Mycelia cultured in SDB for 3 days were also examined, and the results showed that in comparison to the WT (D) and Comp (F) strains, no visible LDs were found in the ΔmrGAT mutant (E) (white arrows point to mitochondria). Bar, 2 μm. (G) Western blot analysis indicated that, in contrast to WT and Comp, cellular accumulation of the LD surface perilipin protein MPL1 was significantly reduced in ΔmrGAT mycelia and conidia.

Fig 5.

Staining with fluorescent dye shows the accumulation of LDs in different cell types of M. robertsii. (A) LD accumulation in the conidia of wild-type (WT), ΔmrGAT, and Comp strains harvested from PDA plates after 20 days. (B) LD distribution in the germlings of WT, ΔmrGAT, and Comp strains grown in a minimum medium with 1% glycerol for 10 h. (C) LD accumulation in the mycelia of WT, ΔmrGAT, and Comp strains grown in SDB for 3 days. (D) LD distribution in appressoria of WT, ΔmrGAT, and Comp strains induced on a hydrophobic surface for 24 h. (E) LD distribution in appressoria of WT, ΔmrGAT, and Comp strains induced on a hydrophobic surface for 48 h. Bar, 5 μm.

mrGAT is required for full virulence in M. robertsii.

We observed that deletion of mrGAT did not impair the formation of appressoria on the surfaces of either hydrophobic plastic plates (Fig. 5D and E) or locust hind wings (see Fig. S6 in the supplemental material). To determine the effect of mrGAT on fungal virulence, we performed injection and immersion bioassays in silkworm larvae. The results demonstrated that the median lethal time (LT₅₀) for the topical infection of the ΔmrGAT mutant (LT₅₀ = 4.30 ± 0.15 days) was significantly longer than for the WT (LT₅₀ = 3.40 ± 0.13 days, χ² = 20.23 and P < 0.0001) and Comp (LT₅₀ = 3.36 ± 0.12 days, χ² = 21.71 and P < 0.0001) (Fig. 6A), which indicated impaired fungal virulence after deletion of mrGAT. However, when the fungal spores were injected directly into the insect hemocoels (body cavities), which bypasses insect cuticles, no significant differences were observed between the ΔmrGAT (LT₅₀ = 1.97 ± 0.04 days) and WT (LT₅₀ = 1.95 ± 0.06 days, χ² = 0.05 and P = 0.8173) strains or between the ΔmrGAT and Comp (LT₅₀ = 1.95 ± 0.04 days, χ² = 0.12 and P = 0.7323) strains (Fig. 6B). These results indicated that deletion of mrGAT reduced virulence by impairing the ability of the fungus to penetrate host cuticle.

Fig 6.

Insect bioassays. (A) Survival of silkworm larvae after topical application of the conidial suspension (1 × 10⁷ conidia ml⁻¹) from WT, ΔmrGAT, and Comp strains. Control insects were treated with 0.05% Tween 20 for 30 s. (B) Survival of silkworm larvae following an injection of 10 μl of 1 × 10⁶ conidia ml⁻¹ suspensions from WT, ΔmrGAT, and Comp strains into the second proleg of larvae. Control insects were injected with 10 μl 0.05% Tween 20.

DISCUSSION

In this study, we present the characterization of mrGAT, which is a GPAT family protein, in a filamentous fungus. We found that in contrast to what occurs in the budding yeast, plant, nematode, and animals, only a single copy of the GPAT gene is present in the genomes of M. robertsii and other insect- and plant-pathogenic fungi belonging to the subphylum Pezizomycotina, Ascomycota. Unlike the lethal effect of GPAT isoform gene deletion in yeasts and nematodes (8, 10), the ΔmrGAT mutant was viable but had reduced abilities to sporulate, accumulate TAG and glycerol, and form LD. The mrGAT null mutant could successfully form appressoria, similar to WT; however, the mutant took a substantially longer time to kill insects than the WT and gene complementation mutant during topical infection but not during injection assays, indicating that the lack of mrGAT diminished fungal capacity to penetrate insect cuticles, which in turn reduced virulence. Deletion of a mammalian DHAPAT-like gene, mrDHAPAT, which is putatively involved in an alternate pathway for LPA production, in M. robertsii did not result in significant changes, if any at all, in mutant physiologies.

The G3P pathway is a crucial physiological process in TAG and phospholipid metabolisms and energy balance (1, 28). In different organisms, various numbers of GPAT isoforms have been reported, and the functions of each isoform in the G3P pathway have been presumed by incorporation of different fatty acid moieties into TAG (3). Different numbers of GPAT are also found in the fungal lineage, with multiple copies observed in basal zygomycete species but only a single copy in most ascomycete species (Fig. 1). This discordance is most likely the result of gene loss that is supported by the absence of GPAT in the lineage of obligate microsporidia, which suffer extensive gene losses during host adaptation (29). However, gene duplication events could not be precluded due to the presence of multiple copies of GPAT in different yeast, mushroom, and zygomycete species (Fig. 1). A single copy of GPAT found in M. robertsii and other fungal species in subphylum Pezizomycotina suggests that in these species the protein functions solely in the G3P pathway. In contrast to the lethal effect of ΔGAT1 ΔGAT2 in yeast mutants (8), the ΔmrGAT mutant is viable but has a reduced yet detectable level of TAG in the mutant cells relative to the control (Fig. 3A). This finding suggests that filamentous fungi could have an alternate pathway for LPA and in turn TAG biosynthesis. Indeed, the presence of the mammalian-protein-like mrDHAPAT in M. robertsii but not in yeast could explain, at least in part, why the ΔmrGAT strain of M. robertsii is viable while ΔGAT1 ΔGAT2 strains of S. cerevisiae cease to grow. The fact that double deletion mutants of mrGAT and mrDHAPAT could not be acquired suggests a similar lethal effect by a complete abolishment of LPA production in the fungus. It remains to be determined whether like yeast GAT1 (7, 8), mrGAT could also convert DHAP into 1-acyl-DHAP or not.

Besides the critical role in initiating TAG biosynthesis, individual GPAT isoforms also contribute to cell polarized growth in yeast (12), LD size increase in fruit fly and mammalian cells (4), and mitochondrial fragmentation in nematode (10). In this study, we found that together with a reduction in cellular TAG level, ΔmrGAT mutants also had impaired sporulation (>90% reduction compared to WT) and LD formation. Fungal conidiation/fertility has been known to be associated with cellular lipid composition and dynamics in M. robertsii and Neurospora crassa (23, 30). Therefore, deletion of mrGAT leading to the failure of fungal sporulation could be due to TAG reduction, which in turn could alter glycerolipid composition. Similarly, deletion of Acl-6 in C. elegans resulted in 70% of the mutants being sterile (10). In D. melanogaster, a DGAT protein (CG8112, isoform A) was also required for oogenesis (31), and functional studies of membrane-bound O-acyltransferases (MBOATs) that contribute to the G3P pathway showed that germ cell development, which is guided by lipid signals in fruit fly, requires redundant protein function (32).

In eukaryotes, TAG is stored as LD in every cell type. Except for the observation that the GPAT4 is associated with LD size growth in flies and mammals (4), deletion of other GPAT isoforms does not directly abolish cellular LD formation in different organisms (1, 3). In this study, we provide the evidence to link GPAT with LD biogenesis in a filamentous fungus in which the number of LDs was significantly reduced in null mutant conidia and completely disappeared in mutant hyphae. LDs are independent organelles that are composed of a neutral lipid core and a phospholipid monolayer anchored by different LD-specific proteins (27). Depending on the cell types, components of the neutral lipid core contain TAG mainly (ca. 44%), DAG (1.6%), cholesteryl esters (ca. 34%), and unknown neutral lipids (ca. 20%) (18). Therefore, the significantly reduced TAG levels in the ΔmrGAT mutant could have contributed to the failure to form the neutral lipid core. In addition, we found that the disruption of mrGAT impaired the accumulation of MPL1 (Fig. 4G), the essential LD surface perilipin protein localized on the phospholipid monolayer to maintain LD structure (17). Taken together, it is not surprising that LD formation is severely impaired or fails in the ΔmrGAT strain.

In both plant- and insect-pathogenic fungi such as M. oryzae and M. robertsii, accumulation of a high concentration of glycerol for building up turgor pressure within the appressorium is a prerequisite for successful penetration of host cuticles (16, 17). In this study, we found that mrGAT but not mrDHAPAT is required for the full virulence of M. robertsii by contributing to fungal penetration of insect cuticles. Along with TAG reduction, glycerol concentration was also reduced in ΔmrGAT cells (Fig. 3B), explaining the loss/reduction of turgor pressure in mutant appressoria. Our data also suggest that similar to what is seen in M. oryzae (19), glycerol production in M. robertsii could be mainly from the lipolysis of TAG rather than from carbohydrate sources. In this respect, G3P production in filamentous fungi could be from the conversion of the glycolytic intermediate dihydroxyacetone phosphate by G3P dehydrogenase (GPD) rather than the phosphorylation of glycerol by glycerol kinase (33). A genome survey found a putative GPD (MAA_06993) in M. robertsii that is similar to the yeast isoforms GPD1 (46% identity) and GPD2 (44%), which control G3P production in S. cerevisiae (34).

In conclusion, our results reveal both the conservative and the divergent roles of GPAT in the G3P pathway in a filamentous fungus model. Unlike what is seen in yeasts, plants, and animals, the single copy of mrGAT and the nonlethal effect of gene deletion in M. robertsii suggest that this fungus can serve as a better model for future studies to elucidate the G3P pathway in fungi, particularly to determine how the G3P pathway is mechanistically connected with LD biogenesis and contributes to fungal pathogenic processes.