Metarhizium robertsii ammonium permeases (MepC and Mep2) contribute to rhizoplane colonization and modulates the transfer of insect derived nitrogen to plants

The endophytic insect pathogenic fungi (EIPF) Metarhizium promotes plant growth through symbiotic association and the transfer of insect-derived nitrogen. However, little is known about the genes involved in this association and the transfer of nitrogen. In this study, we assessed the involvement of six Metarhizium robertsii genes in endophytic, rhizoplane and rhizospheric colonization with barley roots. Two ammonium permeases (MepC and Mep2) and a urease, were selected since homologous genes in arbuscular mycorrhizal fungi were reported to play a pivotal role in nitrogen mobilization during plant root colonization. Three other genes were selected on the basis on RNA-Seq data that showed high expression levels on bean roots, and these encoded a hydrophobin (Hyd3), a subtilisin-like serine protease (Pr1A) and a hypothetical protein. The root colonization assays revealed that the deletion of urease, hydrophobin, subtilisin-like serine protease and hypothetical protein genes had no impact on endophytic, rhizoplane and rhizospheric colonization at 10 or 20 days. However, the deletion of MepC resulted in significantly increased rhizoplane colonization at 10 days whereas ΔMep2 showed increased rhizoplane colonization at 20 days. In addition, the nitrogen transporter mutants also showed significantly higher 15N incorporation of insect derived nitrogen in barley leaves in the presence of nutrients. Insect pathogenesis assay revealed that disruption of MepC, Mep2, urease did not reduce virulence toward insects. The enhanced rhizoplane colonization of ΔMep2 and ΔMepC and insect derived nitrogen transfer to plant hosts suggests the role of MepC and Mep2 in Metarhizium-plant symbiosis.


Introduction
The endophytic insect pathogenic fungus (EIPF) Metarhizium robertsii (Clavicipitaceae) exhibits a varied lifestyle as an entomopathogen, endophyte or as a saprophyte [1]. Metarhizium spp. have been widely used as biocontrol agents against insect pests in agricultural fields. The molecular and biochemical factors regarding insect pathogenicity are relatively well PLOS  studied [2,3]. Several Metarhizium spp. colonize the plant root and form a beneficial symbiotic relationship [1]. However, the genetic factors underpinning plant root or rhizospheric colonization in M. robertsii is largely unexplored. M. robertsii forms a close symbiotic association with certain plants and is capable of transferring nitrogen from infected insects [4,5] to the plant host in exchange for carbon [6]. Relative to the knowledge on the genetic attributes of insect pathogenesis, little is known about the complex relationship between Metarhizium and plant roots. The importance of Metarhizium raffinose transporter (Mrt) [7] and Metarhizium invertase (MrINV) [8] during interaction with plants has been demonstrated. Most plants rely on nitrogen fixing microorganisms or microbial decomposers to acquire nitrogen from the soil. The involvement of nitrogen transporters, ammonium permease and urease, are reported in arbuscular mycorrhizal (AM) fungi during plant colonization [9][10][11][12][13]. The two main nitrogen transporters involved in the nitrogen regulatory network in fungi are ammonium permeases and amino acid permeases [14]. AM fungi acquire nitrogen from the soil in the form of ammonium, nitrate or amino acids, which are then transported to host plants through ammonium/methylammonium permease (AMT/ Mep) transporters. Here inorganic nitrogen or ammonia is assimilated by AM fungi and subsequently converted to arginine in the extraradical mycelium by nitrate reductase or glutamine synthase/glutamate synthase. Arginine is then translocated to the intraradical mycelium where it is converted to urea and ornithine. Urea is then converted to NH 4 + by the action of urease, which is subsequently mobilized to the plant via ammonium transporters/permeases [13]. The symbiotic interaction of Metarhizium with plants [6] suggests the involvement of fungal nitrogen transporters. Nevertheless, genes involved in nitrogen transfer in M. robertsii during plant symbiosis are unknown. Furthermore, Metarhizium share similarities with mycorrhizal fungi in terms of forming symbiotic association and facilitating nutrient exchange with plants, hence it is hypothesized that similar nitrogen transfer mechanisms may be operating in Metarhizium during symbiotic association with plants.
Metarhizium expresses different subsets of genes as a means of physiological adaptation under various conditions. Metarhizium spp. demonstrated large scale differences in gene expression patterns during growth in plant root exudate, insect cuticle or hemolymph as well as in insect host specialization [15][16][17][18]. Although the symbiotic ability and the role as a plant growth promoter has been demonstrated, little is known about the genes involved in this symbiosis. In order to get a better understanding of the genes involved during the symbiotic association of Metarhizium with plants, RNA-Seq was performed on Metarhizium transcripts during bean root colonization and three genes that showed high expression levels were selected for further analysis. A genome survey of M. robertsii revealed the presence of two ammonium permeases (MepC and Mep2) and a urease gene, which has also been selected for the study. Over all, we investigated the involvement of 6 different genes ammonium permease C (MepC; Gene number-MAA_04182), ammonium permease C (Mep2; Gene number-MAA_05002), urease (Gene number-MAA_07458), subtilisin-like serine protease deletion Pr1A (Gene number-MAA_05675), hypothetical protein (Gene number-MAA_08959), hydrophobin 3 (Gene number-MAA_10298) in M. robertsii in barley root colonization and insect pathogenesis by using targeted gene deletions. M. robertsii adhesin 2 deletion (ΔMad2) and a raffinose transporter deletion (ΔMrt) strains were used as comparisons for the plant root colonization assays. In addition, we also quantified the ability of ΔMepC, ΔMep2 and ΔUrease mutants to transfer insect derived nitrogen to barley plants.

Fungal strains and culture conditions
Metarhizium robertsii (ARSEF 2575) wild-type (WT) strain (U.S. Department of Agricultural Research Service Collection of Entomopathogenic Fungal Cultures, Ithaca, NY) was grown and maintained on Potato Dextrose Agar (PDA) (Bioshop Inc.) at 27˚C. M. robertsii ammonium permease C deletion (ΔMepC), ammonium permease 2 deletion (ΔMep2), ΔUrease, subtilisin-like serine protease deletion (ΔPr1A), ΔHypothetical protein (ΔHypo. protein) and hydrophobin 3 deletion (ΔHyd3) strains were used in this study. M. robertsii adhesin 2 deletion (ΔMad2) [19] and a Metarhizium raffinose transporter deletion (ΔMrt) [7] strains were used as comparisons for the plant root colonization assays and were kindly provided by Dr. R.J. St. Leger (University of Maryland). The fungal cultures were routinely grown and maintained on PDA as needed. Conidia were obtained from 12-14 day old PDA plates and conidia were recovered with 0.01% Triton X-100. Conidial yields were then quantified using a hemocytometer and the concentration was adjusted to 1x10 7 conidia/ml.

Targeted gene deletions
Gene deletions based on homologous recombination was conducted according to our previously developed high-throughput methodology [20]. The plasmids for gene deletion were constructed using Gateway Bp Clonase II Enzyme Mix (Invitrogen) or restriction enzyme digestion and ligation. The primers for deletion plasmid construction and other primers used in this study are presented in S1 Table. All PCR products were cloned using high-fidelity Taq DNA polymerase (KOD) and confirmed by DNA sequencing. PCR was performed to verify the targeted gene deletions (S1 Fig).

Growth rate, conidial germination and stress sensitivity assay
The radial growth of M. robertsii WT and gene deletion strains was examined by inoculating the center of the PDA plate with 10 μl of the conidial suspension (1x10 7 conidia/ml), incubated at 27˚C and colony diameter was measured on day 3, 7 and 14. For the stress sensitivity assay, 10 μl of the conidial suspension (1x10 7 conidia/ml) was inoculated on PDA plates supplemented with either 0.01% SDS or 100 μg/ml Congo red. The plates were incubated at 27˚C in the dark and the colony diameters were recorded after 7 days [21]. All assays were performed with 5 biological replicates and the assays were repeated with independent batches of conidial suspension.
The hydrophobicity of ΔHyd3 was assessed using a wettability test [22]. The test was performed by placing 10 μl of a water droplet on the surface of a 14 day old culture and the contact angle of the water droplet was observed after 10 minutes, 30 minutes and 1 hour.

RNA-Sequencing of Metarhizium colonized plant root
Briefly, M. robertsii were allowed to colonize seven day old germinated bean seedlings (Glycine max, Ontario Seed Co Ltd., Ontario, Canada) in sterile soil under greenhouse conditions (25˚C with a photoperiod of 16:8 h light: dark cycle with relative humidity maintained between 60-80%) for 7 days; thereafter the root was harvested, washed, macerated and RNA was extracted using QIAzol lysis reagent (Qiagen) following manufacturer's protocol. Total RNA was extracted from Metarhizium colonized bean roots using QIAzol (Qiagen), following manufacturer's instructions and the concentration of each RNA samples was quantified using Qubit fluorometer (Invitrogen).The purity and the integrity of the RNA samples were assessed spectrophotometrically and agarose gel electrophoresis respectively. The reverse transcription of the sample was prepared and sequenced with the SOLiD TM 4 (Life Technologies) system. One replicate of the Metarhizium colonized bean root was used for RNA-Seq analysis. Metarhizium transcripts were identified and differentiated from plant transcripts by transcript comparison to the Metarhizium genome as the template.
Real-time PCR validation of five Metarhizium genes was conducted to confirm the RNA--Seq results. The primer sequences used for real-time PCR validation are given in S1 Table. Fungal conidial suspension (1x10 6 conidia/ml) was inoculated in potato dextrose broth (PDB) and cultures were incubated at room temperature (22˚C) at 120 r.p.m. Fungal mycelia was collected after 4 days by vacuum filtration and~0.25 g of fungal mycelia was transferred to flasks containing 25 ml of bean root exudate (BRE) and potato dextrose broth (PDB). The haricot bean (Phaseolus vulgaris, Ontario Seed Co Ltd., Ontario, Canada) root exudate was prepared as described [23]. Cultures were then incubated at room temperature (22˚C) at 120 r.p.m and the fungal mycelia was then harvested from bean root exudate after 8 hours of incubation. Total RNA was extracted from fungal mycelia using QIAzol (Qiagen), following manufacturer's instructions. The reverse transcription reactions of each sample were generated using the high capacity cDNA reverse transcription kit (Applied Biosystems) with 4 μg of total RNA in total volume of 80 μl, following manufacturer's recommendations. The transcript levels of each genes were assessed via real-time PCR (BioRad CFX96 real-time PCR system) using Sen-siFast™ SYBR No-ROX kit (Bioline). The real-time PCR reaction mix contained 2 μl of cDNA, 0.4 μl of 10 μM forward and reverse primers, 5 μl of SensiFast™ SYBR No-ROX mix and 2.2 μl of nuclease free water in a final volume of 10 μl. The real-time PCR was performed under following conditions: 95˚C (2 mins), 40 cycles of 95˚C (5 secs), 62-64˚C (10 secs) and 72˚C (20 secs). The specificity of the PCR products was verified by agarose gel electrophoresis. BioRad CFX Maestro (Version: 4.1) was used to analyze the expression of the selected genes using glyceraldehyde-3-phosphate dehydrogenase (gpd) as reference gene. The real-time PCR was performed using three biological replicates and three technical replicates were used for each biological replicate.

Root colonization assay
Root colonization assays were performed using barley (Hordeum vulgare, Sprout Master, Ontario, Canada) as the host plant. The endophytic, rhizoplane and rhizospheric association of WT and gene deletion strains were analyzed. Seeds were surface sterilized with three washes in 4% sodium hypochlorite (NaOCl) for 5 minutes each. The seeds were rinsed with sterile distilled water after each NaOCl wash. The seeds were kept overnight at 4˚C for synchronization of growth before planting. The seeds were then allowed to germinate in water agar (1%) for 3-4 days at 25˚C. The germinated seedlings were then planted in sterile vermiculite (Ther-O-Rock East Inc.). Fungal inoculations were performed using the drench method [24] where 5 ml of the conidial suspension was poured onto the vermiculite surface of each pot. The plants were kept in a greenhouse at 25˚C with a photoperiod of 16:8 h light: dark cycle with relative humidity maintained between 60-80%. The plants were watered daily with sterile distilled water. Five biological replicates were prepared for each treatment. To quantify fungal association, barley roots were harvested from 10 and 20 day old plants. The amount of fungal association on barley roots was analyzed as described previously [23]. To examine the endophytic association, the harvested roots were first washed in water to remove the attached vermiculite. The washed roots were then immersed in 2% NaOCl for 10 seconds and finally rinsed with sterile distilled water to remove traces of NaOCl. The roots were then cut into~2-5mm pieces, weighed and homogenized (Biospec products Inc.) in sterile distilled water for 2 mins. The homogenized root samples were then plated on modified CTC agar (PDA supplemented with 0.5 g/l chloramphenicol, 0.004 g/l thiabendazole and 0.5 g/l cycloheximide) [25] and colony forming units (CFU) values were calculated as CFU/g of root weight. To check the rhizoplane colonization, the harvested roots were treated as described above except the 2% NaOCl treatment was omitted. Water inoculated barley plants were used as controls. Rhizospheric populations of WT and gene deletion strains were also monitored. Here, the vermiculite attached to the barley roots was collected during the harvest, weighed and suspended in 0.01% Triton X-100. The serial dilutions of the suspension were plated on CTC media and the CFU were calculated for per gram of vermiculite.

Insect bioassays
The virulence of the WT M. robertsii and gene deletion strains were assayed against larvae of Tenebrio molitor and Galleria mellonella. An aliquot of 10 μl of the conidial suspension (1x10 7 conidia/ml) was applied to the cuticle of the larvae. Each larva was placed separately in 60 mm x 15 mm diameter Petri dish. Humidity was maintained in each Petri dish with a moistened filter paper. The treated larvae were housed at 25˚C and daily mortality rate was recorded. Each replicate contained 20 larvae and the experiment was duplicated. Controls were 0.01% Triton X-00 treated larvae. The LT 50 values were calculated using Probit analysis.

N-labelled nitrogen transfer assay
The ability of ΔMepC, ΔMep2 and ΔUrease to translocate nitrogen from 15 N -labeled wax moth larvae to the leaves of barley was quantified by the microcosm method as previously described [4,5]. A 10 μl solution of 15 N-labelled ammonium sulphate (5%) was injected through the rear proleg of wax-moth larvae. After 48 hours the larvae were infected with 14 day old fungal conidia. The infected larvae were then placed into microcosms containing autoclaved vermiculite as previously described [4]. The microcosm was then placed in the pots and covered with autoclaved vermiculite and the 3 day old germinated barley seedling was planted on each pot. The 15 N transfer was assessed by the nitrogen transporter mutant strains using two treatments. In the first experimental set, plants were watered daily with sterile distilled water and once a week with 25 ml of 50% Modified Melin-Norkrans (MMN) solution (0.05 g CaCl 2 , 0.025 g NaCl, 0.05 g KH 2 PO 4 , 0.5 g (NH4) 2  In the second treatment, plants were watered daily with sterile distilled water but 50% MMN solution was omitted. The amount of 15 N transfer to plant tissues was determined by harvesting the above ground plant parts after 10 and 20 days. The harvested plant material was dried at 60˚C for 24 hours and were then crushed into a fine powder using a mortar and pestle. The ground plant material was encapsulated in 4-mm by 4-mm tin cups and analyzed for 15 N content by using an NOI-5 emission spectrophotometer.

Nitrogen source assay
The ability of ΔMepC, ΔMep2, ΔUrease and WT to grow on different nitrogen sources was assessed on basal salt agar (BS; 0.1% KH 2 PO 4 , 0.025% Na 2 SO 4 , 0.05% KCl, 0.0125% MgSO 4 �7H 2 O, 0.00625% CaCl 2 , and 1% glucose) [7] supplemented with different nitrogen sources (1 or 30 mM (NH 4 ) 2 SO 4 , 30 mM arginine, 30 mM glutamine, 30 mM glutamate, 1 or 30 mM proline or 30 mM urea). A 1x10 7 conidial/ml suspension (5 μl) was spot inoculated onto BS agar supplemented with different nitrogen sources and growth rates were compared with the WT after 7 days. Ammonium production, based on pH of the gene deletion strains and WT, was also examined. Here, 1 ml of the conidial suspension was added on to potato dextrose broth and allowed to grow for 4 days at 100 rpm at 27˚C. Mycelia was then filtered, washed with sterile distilled water and transferred to minimal media (0.02% KH 2 PO 4 , 0.01% MgSO 4 , 0.2ppm FeSO 4 , 1 ppm ZnSO 4 , 0.02 ppm CuSO 4 , 0.02 ppm MnMoO 4 , 0.02 ppm MnCl 2 ) supplemented with different amino acids (10 mM glutamate, glutamine or arginine). The fungal mycelia (2.5 g) was added to 100 ml of minimal media broth and incubated at 100 rpm at 27˚C. The samples were collected at 0, 6, 24 and 48 hours and the pH were recorded.

Phylogenetic analysis
The amino acid sequences analysis of ammonium permeases of M. robertsii and other fungal ammonium transporters was conducted. Sequence alignments were generated using MUSCLE v3.7 with default parameters [26], and these were then manually refined and end-trimmed to eliminate poor alignments and divergent regions. Unambiguously aligned positions were used for constructing phylogenetic trees with Maximum Likelihood (ML), Bayesian Inference (BI) or distance-based Neighbor-Joining (NJ). A ML tree was constructed using MEGA6.0 (gap treatment: use all sites; 100 bootstrap replications) [27]. The optimal model of phylogenetic relationship was determined using the Find Best Protein Model provided by MEGA 6.0 [27]. A Bayesian inference tree was constructed with MrBayes v3.2.5 as described [28]. The best model was determined as above using MEGA 6.0. For each BI analysis, we used four Metropolis-coupled chains and ran them for 5,000,000 generations, sampling every 1000 generations ('mcmc ngen = 5000000 sample frequency = 1000'). The analysis finished with an average standard deviation of split frequencies of 0.01 or less. The first 25% of trees were discarded as "burn-in". A NJ tree was constructed with default parameters (gap treatment: pairwise deletion; 1000 bootstrap replications) using NJ in MEGA6.0 [27]. Bootstrap support values were obtained by generating 1,000 pseudo-replicates.

RNA-Sequencing and transcriptome analysis
RNA-Sequencing (RNA-Seq) of Metarhizium transcripts during bean root association revealed the expression of over 4000 Metarhizium genes. In the transcriptome data, 217 genes showed relatively higher expression (10-234 readings) and 25% of these upregulated genes were putative uncharacterized or hypothetical protein. Of these, the top five most highly expressed M. robertsii genes (~90-234 readings) were hydrophobin (MAA_10298), tubulin beta chain (MAA_02081), subtilisin-like serine protease Pr1A (MAA_05675) and two other genes that encode putative uncharacterized proteins (MAA_08959 and MAA_00771). The protein sequence of MAA_08959 (hypothetical protein) was surveyed in GenBank, and similar sequences were reported in other Metarhizium species, however these were also categorized as putative uncharacterized or hypothetical protein. A homolog of MAA_08959 was reported in M. brunneum as filamin/ABP280 repeat like protein but was not functionally characterized. S2 Table shows

Phylogenetic analysis
BLAST search using the NCBI protein database of MepC showed amino acid sequence similarities to the ammonium transporters of phytopathogens and other insect pathogens/plant root colonizers, with more than 80% similarity to other Hypocreales such as Ustilaginoidea virens, Beauveria bassiana and Trichoderma harzianum (Fig 1). These fungi represent diverse lifestyles as phytopathogens, endophytic insect pathogens and endophytic mycoparasites, respectively. Phylogenetic analysis showed that Mep2 amino acid sequences clustered with ammonium permeases from phytopathogenic and endophytic fungi. Mep2 showed a close The BLAST analysis of urease showed more than 80% amino acid sequence similarities with Purpureocillium lilacinum (entomopathogenic, rhizospheric), Tolypocladium capitatum (insect parasitic, mycoparasitic), Hirsutella minnesotensis (insect and nematode pathogenic) and other endophytic/phytopathogenic fungi (Trichoderma spp., Colletrotrichum spp., Verticillium spp. and Fusarium spp.). The nucleic acid sequence analysis of MAA_10298 revealed sequence identity with a functionally characterized Class I hydrophobin from M. brunneum, HYD3, and hence named as Hyd3 in this study.

Effect of gene disruption on phenotypic characteristics
The growth rate of WT M. robertsii and the gene deletion strains were similar on PDA at 27˚C (Fig 2A). Disruption of Hyd3 affected conidiation on PDA and conidial yield was decreased bỹ 7 fold compared with WT (ANOVA, p<0.001) (Fig 2B). No significant differences in conidiation were noted for other gene deletion strains relative to WT (Fig 2B). However, differences in the colony morphology were noted in gene deletion strains relative to WT. The ΔHypo. protein showed a thick white cottony mycelium (Fig 2C) compared to WT on PDA. The effect of various stress conditions on the growth of gene deletion strains was assayed on agar plates containing either Congo red or SDS. No significant differences in the growth rate were observed in any of the stress conditions (Fig 2C). The hydrophobicity test showed that ΔHyd3 was a wettable phenotype observed as a decrease in the contact angle of the water droplet on the colony surface (Fig 2D). The WT retained the water droplet after 1 hour while ΔHyd3 exhibited a less hydrophobic colony surface.

Effect of gene disruption on rhizoplane and endophytic colonization ability
The ammonium permease deletions, ΔMepC and ΔMep2 showed increased rhizoplane colonization after 10 or 20 days respectively compared to the WT (Fig 4A). A significant increase in CFU was observed for ΔMepC (24.5x10 4 CFU/g of root) at 10 days post treatment as compared to WT (7.1x10 4 CFU/g of root) (Tukeys multiple comparison test: WT v/s ΔMepC (p<0.001)). However, ΔMepC showed no significant difference in rhizoplane colonization levels relative to WT after 20 days. In contrast, ΔMep2 showed similar levels of root colonization as the WT after 10 days; however, significantly higher CFU levels (35.5 x10 4 CFU/g of root) was recovered after 20 days (WT v/s ΔMep2 (p<0.0001). The sugar transporter deletion ΔMrt showed a similar rhizoplane colonization pattern as ΔMepC. Rhizoplane colonization for ΔMrt (27.7x10 4 CFU/g of root) after 10 days post treatment was ca. 4 times greater than WT. Nevertheless, the recovery of ΔMrt (4.3x10 4 CFU/g of root) from plant roots after 20 days post inoculation was similar to the WT (3.6x10 4 CFU/g of root). The disruption of the urease gene in M. robertsii had little impact on root colonization. Similarly, the disruption of Hyd3, Pr1A and Hypo. protein did not affect the rhizospheric or rhizoplane colonization abilities compared with the WT. When the rhizospheric persistence of these strains in vermiculite was analyzed, there were no significant differences in the CFU recovered from vermiculite compared to WT except for ΔMepC at day 10 ( Fig 4B). No significant differences in endophytic colonization were observed for the gene deletion strains compared with the WT (S3 Table).

Insect derived 15 N transfer to barley by Metarhizium nitrogen transporter mutants
Metarhizium strains including WT and the nitrogen transporter mutants were able to transfer significant amounts of insect derived-15 N to barley after 10 and 20 days of growth (Fig 5). M. robertsii WT showed significant difference in the insect derived-15 N incorporation between MMN treated and untreated plants after 10 days of growth in microcosms. A significantly lower 15 N incorporation was noted in MMN treated plants (68.84%) versus untreated plants (84.33%) when grown in microcosms containing M. robertsii WT infected, 15 N injected wax moth larvae. However, no significant difference was observed in insect derived-15 N incorporation between MMN treated and untreated after 20 days of growth in the presence of WT infected, 15 N-injected wax moth larvae.
All three of the nitrogen transporter mutants showed significantly higher 15  ΔMepC (79.75%) showed significantly less 15 N transfer compared to WT (84.33%) (Tukeys multiple comparison test: WT v/s ΔMepC (p<0.03)). However, no significant difference in insect derived-15 N incorporation by ΔMepC was observed relative to WT in both MMN treated and untreated plants after 20 days of growth.
In the MMN treated barley plants, ΔMep2 showed significantly greater 15 N transfer to barley for both 10 (78.10%) and 20 (82.95%) days compared to WT (68.84% and 77.48% of 15 N incorporation for 10 and 20 days respectively) (WT v/s ΔMep2 (p<0.005)). However, no significant difference in insect derived 15 N incorporation relative to WT was noted in plants that were not treated with MMN for both 10 and 20 days. ΔUrease, showed no significant difference relative to WT in all the experimental condition except 10 days in MMN treated plants. A significantly higher 15 N incorporation to barley was noted in the presence of ΔUrease (79.93%) compared to WT (68.84%) was noted after 10 days in MMN treated plants (WT v/s ΔUrease (p<0.0001)).

Nitrogen source assay
The influence of various nitrogen sources on the colony diameter (Fig 6) ΔMep2) and ΔUrease when growth in BS agar plates supplemented with either arginine or glutamine (30 mM) as the sole nitrogen source. In 1 mM proline, ΔMepC showed reduced colony diameter, while the colony diameter was similar to the WT in 30 mM proline. Interestingly, ΔMep2 exhibited a colony diameter similar to the WT in 1 mM proline, whereas a significantly higher colony diameter than WT was noted in 30 mM proline. ΔUrease showed less growth in BS agar medium supplemented with urea as a sole nitrogen source (S3 Fig). No significant differences were observed in pH after growing in YPD and transferred to minimal media containing arginine. A slight increase in pH was noted for the WT after 6 hours in glutamine compared to the mutant strains. Additionally, ΔMep2 showed an increase in pH relative to WT after 6 hours and 24 hours when transferred to minimal media containing glutamate (S4 Fig).

Discussion
Here we report the effects of deletions of two ammonium transporters in M. robertsii, MepC and Mep2 and a sugar transporter Mrt on barley root rhizoplane association and the transfer of insect derived nitrogen to plant hosts. The targeted deletion of a urease gene in M. robertsii, however showed no effect on root colonization. The gene deletions, ΔPr1A, ΔHypo. protein and ΔHyd3, that were chosen based on the high level of transcripts during root colonization showed no differences in root colonization compared with the WT.
Nitrogen transfer to plants during symbiosis is a common feature described in arbuscular mycorrhizal fungi which is mediated via ammonium permeases [13]. Our study showed that the targeted deletion of MepC or Mep2 in M. robertsii resulted in enhanced rhizoplane colonization at 10 and 20 days respectively, while the colonization efficiency of ΔUrease was unchanged compared to the WT. Ammonium and glutamine are preferred nitrogen sources for fungi. When these nitrogen sources are limited, fungi utilize other nitrogen sources. The regulatory system that enables the selective utilization of secondary nitrogen sources is known as nitrogen metabolite repression or nitrogen catabolite repression [29]. This process allows for the adaptability of fungi to changing nitrogen sources. The expression of AMT/Mep were subjected to nitrogen metabolite repression and this feature was described in non-pathogenic filamentous fungus, Aspergillus nidulans [30], phytopathogenic fungi (Colletotrichum gloeosporioides) [31] ectomycorrhizal fungi (Hebeloma cylindrosporum) [32]. The two GATA transcription factors identified in mediating nitrogen catabolite repression are AreA (A. nidulans) and nit-2 (N. crassa) [33]. The functionally equivalent homologues of these regulators known as nitrogen response regulator gene (nrr1) are reported in Metarhizium and are suggested to play a critical role in insect virulence by regulating subtilisin-like protease and trypsin like protease expression [34]. Similar mechanisms may be operating in Metarhizium for plant colonization. The deletion of the ammonium permease gene may have triggered secondary nitrogen utilization pathways which subsequently increased root colonization compared with the WT. The nitrogen source growth assays revealed the differential growth responses of ΔMep2 and ΔMepC in varying nitrogen conditions. ΔMepC showed reduced growth under low nitrogen conditions, while ΔMep2 showed no growth impairment. Previous studies demonstrated the differential expression of mepA and mepB in A. nidulans which were regulated by the transcription factor AreA [33]. The differential activation of ammonium transporters in response to ammonium and, consequentially, appressoria formation and virulence has been reported in the phytopathogenic fungus C. gloeosporioides. Their data suggested that the balance between ammonium uptake and release can induce appressorium formation by triggering signaling pathways [31]. Studies have also reported the role of signaling pathways including MAP kinase and cAMP in regulating appressorium formation and virulence in phytopathogenic [35] and entomopathogenic fungi [36,37]. This suggests the role of signaling pathways in sensing host related stimuli (i.e. the form of nitrogen) in initiating fungal colonization on hosts. This phenomenon is not restricted to nitrogen but also to carbon source. For example, M. robertsii ΔMrINV resulted in a severe reduction of growth in vitro in root exudate. Conversely it showed enhanced root colonization ability on both switch grass and Arabidopsis. The increased root colonization was suggested as catabolite "derepression" in the invertase deletion [8]. Interestingly, no significant differences in endophytic colonization were noted for the nitrogen transporter, ΔMrt or the adhesin gene ΔMad2 deletions at 10 and 20 days.
Previous studies has shown that several species of Metarhizium including generalist and specialist insect pathogens were able to transfer insect derived nitrogen to plant hosts [4,5]. The 15 N-transfer assay showed that, the three nitrogen transporter mutants (ΔMepC, ΔMep2 and ΔUrease) showed greater incorporation of insect derived 15 N to barley plants compared to WT when treated with 50% MMN. This suggests the adaptability of Metarhizium to survive in the nitrogen rich environment in the absence of major nitrogen transporters by triggering secondary nitrogen utilization pathways. We observed a significantly low 15 N incorporation by ΔMepC relative to WT in the absence of 50% MMN after 10 days of growth in microcosm. In contrast, the plant colonization assay revealed greater colonization by ΔMepC relative to WT after 10 days of association. We speculate that this variation is due to the difference in the nutrient sources available to the fungi in the rhizospheric environment that trigger different compensatory pathways dependent on nutrient availability. Previous studies reported the expression of different subset of genes by Metarhizium when colonizing different hosts or in different physiological conditions [15]. Furthermore, studies have shown that the pH in the rhizospheric environment can influence the amount of nitrogen absorbed by the plants [38]. Previous studies demonstrated the significance of pH as an environmental cue in the host niche as a differential gene expression determinant [39].
Fang and St Leger [7], reported that the disruption of the raffinose transporter gene (ΔMrt) greatly reduced the rhizosphere competency of M. robertsii after 2 and 3 months post inoculation while no significant differences in rhizospheric populations were noted with WT and ΔMrt after 1 week or 1 month [7]. Consistent with this, we found that ΔMrt populations in the rhizosphere remain constant and did not significantly differ from the WT after 10 or 20 days. Surprisingly, we observed enhanced rhizoplane colonization for ΔMrt strains compared to WT at 10 days similar to ΔMepC. In contrast, a previous study has shown that ΔMrt was able to colonize corn roots similar to WT but the population in the rhizospheric soil was reduced due to the inability of ΔMrt to uptake oligosaccharides from root exudate [7,40]. The cell wall protein Mad2 is a specific adhesin which enables M. robertsii to adhere to plant roots. Field trial experiments using corn demonstrated a reduction in survival of ΔMad2 in rhizospheric soil as well as reduced ability to colonize plant roots [40]. However, in our study, we did not observe any differences in rhizoplane or rhizospheric association for ΔMad2 compared to the WT. This inconsistency in ΔMrt and ΔMad2 with regard to root colonization ability could be probably due to the difference in the fungal inoculation method (drench method around planted seedlings) as well as the substrate (vermiculite) we employed for planting, while Metarhizium treated corn seeds were used for the field trial study [40].
The targeted deletion of genes (Pr1A, Hyd3 and Hypo. protein) that were upregulated in plant roots (RNA-Seq data) had no impact on endophytic, rhizoplane or rhizospheric colonization. The deletion of subtilisin-like serine protease (Pr1A) did not affect root colonization ability of M. robertsii. Pr1A is the predominant protease upregulated during Metarhizium infection of the insect cuticle [41] and is an important gene used for strain improvement for biocontrol purposes. The integration of multiple copies of Pr1A and its constitutive expression in M. anisopliae increased virulence against insect hosts [42]. The expression of subtilisin-like protease has been reported in plant pathogenic fungi, Magnaporthe poae, a pathogen of Kentucky bluegrass. The expression of the protease increased on infected roots [43]. Similar subtilisin-like proteases have also been demonstrated in grass endophytic fungi including Neotyphodium typhinum [44] and Epichloe festucae [45], however, the specific role of these proteases during plant colonization is unknown. Greater expression levels of Pr1A has been demonstrated in M. robertsii grown in bean root exudate [46], however, evidence suggests that the expression of Pr1A is correlated to a stress response and nutrient availability. Hence, the expression of Pr1A is proposed as a stress response by M. robertsii under nutrient starvation conditions including bean root exudate [15] and possibily during plant root colonization. Moreover, previous studies suggested the evolutionary role of different proteases that resulted from gene duplication, loss and horizontal gene transfer events, is correlated with the multiple life styles of M. robertsii [18].
The role of hydrophobins during initial interaction and virulence with plant hosts have been investigated in phytopathogenic and endophytic fungi [23,[47][48][49][50]. Although Hyd3 was upregulated in Metarhizium-bean root transcriptome, the disruption of Hyd3 did not affect the rhizoplane or rhizospheric colonization ability of M. robertsii. Sequence analysis revealed 96% sequence similarity with M. brunneum HYD3. However, the loss of the Hyd3 gene did not affect plant rhizosphere competency for M. brunneum (unpublished data). Furthermore, previous studies have also revealed the contribution of hydrophobins in the entomopathogenic/ endophytic fungus B. bassiana in root colonization of Phaseolus vulgaris [23]. The consequences of the targeted deletion of hydrophobin genes on fungal development and interaction with hosts can vary in different fungi. Besides, the presence or upregulation of other hydrophobins or signaling pathways could compensate or mask homologous gene deletions.
M. robertsii can infect and kill a wide range of insect species [51]. The molecular and the biochemical factors involved in insect pathogenicity is well studied [2,52]. Disruption of MepC, Mep2, urease or Hypo. protein gene did not alter the virulence of M. robertsii against mealworm or wax-moth larvae. However, the disruption of the Hyd3 or Pr1A resulted in delayed mortality compared to the WT but only against mealworm larvae. The deletion of Hyd3, which encodes a Class I hydrophobin, resulted in altered hydrophobicity and conidiation in M. robertsii. Consistent with this, the deletion of HYD3 affected conidiation, surface hydrophobicity and pathogenicity in M. brunneum [22]. In B. bassiana, two Class I hydrophobins, hyd1 and hyd2 were reported to play distinct roles in fungal development and interaction with insect hosts. The targeted deletion of hyd1 significantly altered rodlets on the surface of conidia, lowered surface hydrophobicity and virulence, but the conidia retained adhesion qualities. While the deletion of hyd2 resulted in decreased surface hydrophobicity and adhesion phenotype but the virulence towards the insect host was not affected [21]. Subtilisin-like serine proteases have been reported in nematode-parasitic, mycoparasitic and entomopathogenic fungi as a virulence factor [53,54]. Subtilisin-like proteases can degrade the protein linkages present on the host integument and thus mediates the penetration and further colonization events during infection. The role of M. robertsii Pr1A during insect pathogenesis has been reported previously [55].
In conclusion, although high levels of expression of Pr1A, Hypo. protein, Hyd3 was observed in bean root, the disruption of these genes did not affect the root or rhizosphere colonization ability of the fungus. The deletion of Hypo. protein had little impact on insect virulence, while Pr1A and Hyd3 contributed to pathogenicity against meal worm. The expression of Hyd3 plays a significant role in conidial hydrophobicity in M. robertsii. The impact of the gene deletions depends on several features including compensatory gene expression or the plasticity adapted by the organism in order to survive in that environment. The effect of synthetic gene alterations may be effective only under certain environmental or nutrient conditions where the organism is limited in its ability to express compensatory genes [56]. Furthermore, we report the involvement of ammonium permease genes (MepC and Mep2) of M. robertsii during plant root colonization as well as in insect derived nitrogen transfer. Nitrogen transporters, MepC, Mep2 and urease have no role in insect pathogenicity which was confirmed in two insect hosts. These genes play a critical role in the growth of fungi in nitrogen rich and low conditions. However, a deeper insight into the underlying processes that results the stable association between M. robertsii and plant host are necessary. The identification and characterization of symbiosis related genes and specific functions will provide a deeper understanding on plant root colonization formed between this ecologically and agronomically important fungus and different plant hosts.
Supporting information S1 Table. Primer pairs used in the study.