A High-Throughput Gene Disruption Methodology for the Entomopathogenic Fungus Metarhizium robertsii

Systematic gene disruption is a direct way to interrogate a fungal genome to functionally characterize the full suite of genes involved in various biological processes. Metarhizium robertsii is extraordinarily versatile, and it is a pathogen of arthropods, a saprophyte and a beneficial colonizer of rhizospheres. Thus, M. robertsii can be used as a representative to simultaneously study several major lifestyles that are not shared by the “model” fungi Saccharomyces cerevisiae and Neurospora crassa; a systematic genetic analysis of M. robertsii will benefit studies in other fungi. In order to systematically disrupt genes in M. robertsii, we developed a high-throughput gene disruption methodology, which includes two technologies. One is the modified OSCAR-based, high-throughput construction of gene disruption plasmids. This technology involves two donor plasmids (pA-Bar-OSCAR with the herbicide resistance genes Bar and pA-Sur-OSCAR with another herbicide resistance gene Sur) and a recipient binary plasmid pPK2-OSCAR-GFP that was produced by replacing the Bar cassette in pPK2-bar-GFP with a ccdB cassette and recombination recognition sites. Using this technology, a gene disruption plasmid can be constructed in one cloning step in two days. The other is a highly efficient gene disruption technology based on homologous recombination using a Ku70 deletion mutant (ΔMrKu70) as the recipient strain. The deletion of MrKu70, a gene encoding a key component involved in nonhomologous end-joining DNA repair in fungi, dramatically increases the gene disruption efficiency. The frequency of disrupting the conidiation-associated gene Cag8 in ΔMrKu70 was 93% compared to 7% in the wild-type strain. Since ΔMrKu70 is not different from the wild-type strain in development, pathogenicity and tolerance to various abiotic stresses, it can be used as a recipient strain for a systematic gene disruption project to characterize the whole suite of genes involved in the biological processes of M. robertsii.


Introduction
The genus Metarhizium includes the best-studied entomopathogenic fungi at the molecular and biochemical levels. They have a worldwide distribution from the arctic to the tropics and colonize an impressive array of environments including forests, savannahs, swamps, coastal zones and desserts [1]. The representative species of the genus, Metarhizium robertsii, is extraordinarily versatile. This versatility derives from the fact that it is a pathogen of arthropods [1] and a saprophyte [2], and some isolates such as ARSEF2575 (used in this study) are the beneficial colonizers of rhizospheres (the layer of soil influenced by root metabolism) [3,4]. Consequently, M. robertsii experiences several distinctive sets of selective pressures [5,6], allowing the function of a gene to be assessed during the three major fungal lifestyles including parasitism, saprophytism and symbiotism. This is not matched by the ''model'' fungi Saccharomyces cerevisiae and Neurospora crassa [7].
Systematic gene disruption is a direct way to interrogate a fungal genome to functionally characterize the full suite of genes involved in various developmental processes. The completion of a systematic gene disruption project for a fungus is dependent on two important technologies. The first is a highly efficient gene disruption technology based on homologous recombination, and this technology is particularly important for filamentous fungi because they usually have a very low homologous recombination frequency [8,9]. This is because filamentous fungi preferentially use nonhomologous end-joining DNA repair (NHEJ) in doublestrand break (DSB) repair. As a result, exogenous DNA can be randomly integrated in the genome, even if it carries a long stretch of homologous sequence [8]. NHEJ is mediated by the DNAdependent protein kinase catalytic subunit (DNA-PKcs), the Ku70-Ku80 heterodimer, and the DNA ligase IV-Xrcc4 complex. Deletion of the genes encoding Ku proteins impairs NHEJ (i.e., decreases the rate of non-homologous recombination), and thus significantly increases the rate of homologous recombination in the filamentous fungi [10]. Furthermore, the deletion of Ku genes does not alter their growth, development and infection [8,9,11,12]. Therefore, mutants containing Ku deletions are used as recipient strains for systematic gene disruption in N. crassa and also for functionally characterizing genes involved in various biological processes in other fungi [8,9,11,12]. The second technology is high-throughput approaches to generate gene disruption constructs. Such approaches include double-joint PCR fusion in the PCR-based, split-marker deletion method [13] and OSCARbased (One Step Construction of Agrobacterium Recombination ready plasmids) construction of gene disruption plasmids [14]. Protoplast preparation and transformation are needed for the PCR-based split-marker deletion method, and A. tumefaciensmediated fungal transformation is needed for the OSCAR-based technology.
The PCR-based, split-marker deletion method appears not to be suitable for M. robertsii mainly because the efficiency of protoplast-mediated M. robertsii transformation is very low [15]. In contrast, A. tumefaciens-mediated M. robertsii transformation is highly efficient. The current strategy for constructing plasmids for A. tumefaciens-mediated gene disruption in M. robertsii requires at least two rounds of restriction enzyme digestion and ligation, which is labor intensive and thus not feasible for high-throughput gene deletion. In this study, we describe a modified OSCAR methodology that facilitates the fast and high-throughput generation of gene disruption plasmids for M. robertsii. Disruption of the M. robertsii gene MrKu70 (homologous to Ku70 of N. crassa) showed that the efficiency of gene disruption based on homologous recombination in DMrKu70 was much higher than that in the wild-type strain.

Modified OSCAR-based technology for high-throughput construction of gene disruption plasmids
In 2011, Paz et al. described an OSCAR technology that was a combination of PCR and the Gateway Technology and could generate a gene disruption plasmid with a single cloning step in just 2 days [14]. In this technology, the flanking region of the gene to be deleted are cloned by PCR and recombined into a donor plasmid (e.g., pA-Hyg-OSCAR) and a recipient binary plasmid (p-OSCAR) catalyzed by Bp Clonase (Life Technologies, USA), which generates the resultant gene disruption plasmid [14]. In order to use the OSCAR technology in M. robertsii, we modified both donor plasmids and the recipient binary plasmid. The hygromycin-resistance gene in pA-Hyg-OSCAR and nourseothricin-resistance gene in pA-NTC-OSCAR constructed by Paz et al. [14] are effective for Verticillium dahlia and many other fungi, but are not suitable for M. robertsii. Therefore, we replaced the HygR cassette in pA-Hyg-OSCAR with either of the two herbicideresistance gene cassettes Bar and Sur, generating two new donor plasmids referred to as pA-Bar-OSCAR ( Fig. 1) and pA-Sur-OSCAR ( Fig. 1), respectively. To produce a recipient binary plasmid suitable for M. robertsii, the Bar cassette in the plasmid pPK2-bar-GFP [3] was replaced by the DNA fragment containing the recombination recognition sites (attp3 and attp2r) and the ccdB cassette from the plasmid p-OSCAR [14]. The new recipient binary plasmid is designated as pPK2-OSCAR-GFP (Fig. 1). The green fluorescent protein gene egfp in the plasmid can be used to aid in screening gene disruption mutants as described [16]. Taken together, we developed a modified OSCAR technology that contains two donor plasmids (pA-Bar-OSCAR and pA-Sur-OSCAR) and one recipient binary plasmid pPK2-OSCAR-GFP. The process of constructing gene disruption plasmids using these plasmids is outlined in Figure 1. Using this modified OSCAR technology, we successfully constructed many plasmids for disrupting M. robertsii genes. In this study, we describe the disruption of the conidiation-associated gene Cag8 and the M. robertsii MrKu70 gene, which is homologous to Ku70 of N. crassa and is involved in NHEJ [8].
Characterization of the Ku70 gene in M. robertsii MrKu70 (Genbank accession number: XP_007817332) was identified by BLASTP using the Ku70 protein of N. crassa [8] as a query. MrKu70 consists of 646 amino acids containing a von Willebrand factor type A (vWFA)-like domain, a Ku-core domain (Ku70), and a DNA-binding domain (SAP domain). This domain structure is conserved in all characterized Ku70 orthologs of eukaryotic origin [8,9,11,12]. MrKu70 exhibits 87, 82, 77, and 77% similarity to its orthologs found in Trichoderma reesei (ETS02014), Trichoderma virens (EHK20312), N. crassa (Q7SA95) and Sordaria macrospora (XP_003350038), respectively. We then constructed an MrKu70 disruption plasmid using the modified OSCAR methodology with Sur as the selection marker, and disrupted MrKu70 based on homologous recombination. Six disruption mutants (DMrKu70) were obtained from 200 transformants with chlorimuron ethyl resistance. DMrKu70 was complemented by a genomic clone of MrKu70. The details of the disruption of MrKu70 and the complementation of DMrKu70 are described in Figure 2.
We tested the tolerance of DMrKu70 conidia (two different mutants) to several abiotic stresses including UV-B, heat shock (45uC for 2 h), cold (15uC), high osmolarity (0.75 M KCl) and an oxidant (0.005% H 2 O 2 ). No significant differences in tolerance to the abiotic stresses were observed between DMrKu70, the wildtype strain and the complemented DMrKu70 (P.0.1) (data not shown). Bioassays with wax worm larvae showed the LT 50 (the time taken to kill 50% of G. mellonella larvae) of the wild-type strain was 10.861.6 days, which was not significantly different from that of DMrKu70 (11.061.9 days) and the complemented DMrKu70 (11.161.2 days) (P.0.05). These results indicate that the disruption of MrKu70 did not alter the pathogenicity of M. robertsii to this insect. DMrKu70 also showed wild-type levels of germination in Sabouraud Dextrose Broth, growth and conidiation on Potato Dextrose Agar (PDA, Becton Dickinson and Company, France) (data not shown). In order to test the stability of the biological processes of two DMrKu70, the wild-type strain and the complemented DMrKu70, they were successively subcultured on PDA plates five times. Conidia from each subculture were collected and subjected to the same tests described above. No significant differences in the biological processes tested were found between subcultures of DMrKu70, the wild-type strain and the complemented DMrKu70 (P.0.05). These data suggest that DMrKu70 can be used as a recipient strain for gene disruption to functionally characterize genes involved in pathogenesis, development or tolerance to abiotic stresses in M. robertsii.

Deletion of MrKu70 increased gene disruption efficiency in M. robertsii
In order to test the gene disruption efficiency in DMrKu70, a well-defined Cag8 gene in M. robertsii was selected as a target for disruption. The deletion of Cag8 resulted in a fluffy phenotype without conidiation [17], and this visible phenotype change can allow us to readily evaluate the efficiency of gene disruption (Fig. 3B). The disruption plasmid for Cag8 was generated using the modified OSCAR methodology with Bar as the selection marker, and the 59 and 39 flanking regions were 994 and 1,286 bp, respectively. Only 21 out of 325 glufosinate ammonium-resistant transformants (6.5%) from the wild-type strain were fluffy without conidia, while 281 out of 301 transformants (93.4%) from DMrKu70 were fluffy without conidia (Fig. 3B). PCR confirmation showed that Cag8 was successfully disrupted in two randomly selected fluffy transformants (Fig. 3C). The complementation of DCag8 was not conducted because Cag8 had been previously characterized [17]. These data indicate the disruption of MrKu70

Discussion
Systematic gene disruption projects have been completed in the two 'model' ascomycete fungi S. cerevisiae (http://www-sequence. stanford.edu/group/yeast_deletion_project/deletions3.html) and N. crassa (http://www.dartmouth.edu/,neurosporagenome/ proj_overview.html), and research based on the two projects has been of extraordinary utility in advancing our knowledge of fundamental principles of biology. However, the fungal kingdom, with an estimated 1.5 million different species [18], displays extraordinary evolutionary diversity, and model fungi such as S. cerevisiae and N. crassa cannot be used to elucidate the mechanisms involved in major fungal lifestyles that they do not share [7].
Metarhizium robertsii can be used as a representative to simultaneously study several major lifestyles that are not shared by the ''model'' fungi S. cerevisiae and N. crassa; a systematic examination of gene function in this fungus will benefit studies on other fungi. Therefore, we developed a high-throughput gene disruption methodology for M. robertsii, which includes two technologies. One is the modified OSCAR-based high-throughput construction of gene disruption plasmids. Using this technology, a gene disruption plasmid can be constructed in one cloning step in two days. The other technology is highly efficient gene disruption in the MrKu70 deletion mutant. Since the disruption of MrKu70 does not alter development, pathogenicity and tolerance to various abiotic stresses, DMrKu70 can be used as a recipient strain for systematic gene disruption to characterize the full suite of genes involved in the biological processes of M. robertsii. However, the insect species (G. mellonella) used in this study is not a natural host of M. robertsii; therefore, so bioassays of DMrKu70 should be performed when a new insect species is used. Furthermore, we propose that the wild-type strain and DMrKu70 should always be used as controls to investigate the function of any gene using the gene disruption methodology. In addition, the deletion of Ku70 genes could predispose fungal strains to DNA damaging conditions [11,[19][20][21]. Therefore, DMrKu70 could been not suitable for studying genes involved in biological processes related to an intact NHEJ pathway.
In N. crassa, the gene deletion frequency using NHEJ-defective strains increases by increasing the size of flanking regions in the gene replacement cassette, and the gene deletion frequency Figure 1. The modified OSCAR methodology for construction of gene disruption plasmids. The donor plasmids contain the cassette of the herbicide-resistance genes Bar or Sur. The recipient binary plasmid pPk2-OSCAR-GFP was used for construction of gene disruption plasmids in E. coli. Primers B2r+5F/B1r+5R and B4+3F/B3+3R were used for cloning 59 and 39 flanking sequences of the gene to be deleted, respectively. B2r, B1r, B4 and B3 are recognition sites of Bp Clonase for recombination, which were added to 59 end of their respective primers. Primer 5F (or B2r+5F) and Bar3 (Sur3 when pA-Sur-OSCAR used) and Bar5 (Sur5 when pA-Sur-OSCAR used) and 3R (or B3+3R) were used to confirm the construction of the gene disruption plasmids. doi:10.1371/journal.pone.0107657.g001 reached 100% when the size was 1,000 bp. Therefore, approximately 1,000 bp of flanking regions were first used to construct the Cag8 disruption cassette, and the gene disruption frequency was 93%. Increasing the size of the flanking regions might promote gene deletion frequency in DMrKu70; however, this was not carried out because a gene disruption frequency over 90% with an efficient fungal transformation system is sufficient for highthroughput gene deletion.
The herbicide resistance genes Bar or Sur are suitable for many other fungi such as the entomopathogenic fungus Beauveria bassiana [22] and the plant pathogenic fungus Magnaporthe oryzae [23]. Therefore, the plasmids developed in this study (pAbar-OSCAR, pA-sur-OSCAR and pPK2-OSCAR-GFP) can be directly used for high-throughput construction of gene disruption plasmids for these fungi, possibly facilitating their functional genomic analysis.

Fungal and bacterial strains
Metarhizium robertsii ARSEF2575 was cultured as previously described [17]. E. coli DH5a and DB101 were used for plasmid construction. A. tumefaciens AGL1 was used for fungal transformation.

Constructing OSCAR plasmids for gene disruption
In order to replace the hygromycin-resistance gene cassette (HygR) in pA-Hyg-OSCAR [14] with either of the two herbicide resistance gene cassettes (Bar and Sur), pA-Hyg-OSCAR was first digested with PstI and blunted with T4 DNA polymerase. The linear plasmid was then digested with SpeI to remove the hygromycin-resistance gene cassette. The Bar cassette (cut off with SpeI/EcoRV from pPK2-bar-GFP) [24] and the Sur cassette (cut off with XbaI/EcoRV from pPK2-SUR-GFP) [24] were ligated with the linearized pA-Hyg-OSCAR to produce pA-Bar-OSCAR and pA-Sur-OSCAR, respectively. All restriction enzymes and DNA modification enzymes used in this study were from Fermentas (Beijing, China).
The recipient binary plasmid pPK2-OSCAR-GFP was constructed by ligating pPK2-Bar-GFP (Bar cassette deleted) with the DNA fragment containing a ccdB cassette and Bp Clonase recognition sites (attp3 and attp2r). To delete the Bar cassette in pPK2-Bar-GFP, the binary plasmid was digested with EcoRI/ EcoRV and blunted with T4 DNA polymerase. The DNA fragment containing the ccdB cassette and Bp Clonase recognition sites was removed from p-OSCAR [14] with KpnI/HindIII and blunted with T4 DNA polymerase.

Gene disruption
The gene disruption plasmids were constructed as described by Paz et al. [14]. For each gene, two pairs of primers were designed to amplify approximately 1 Kb of the 59 and 39 flanking sequences, respectively, and each primer had a unique attB sequence at its 59end. The details of all primers used in this study are described in Table 1. The PCR products, the donor plasmid (p-Bar-OSCAR or p-Sur-OSCAR), the recipient plasmid (pPK2-OSCAR-GFP) and BP Clonase (Life Technologies, Shanghai China) were prepared as described by Paz et al. [14]. After a 16-h incubation at 25uC, the mixture was transformed into E. coli (DH5a) mediated by CaCl 2 [25]. Confirmation of construction of gene disruption plasmids is described in Figure 1.
To complement DMrKu70, a genomic DNA fragment of MrKu70 including the promoter region, ORF and termination region was cloned by PCR using primers MrKu70-5 and MrKu70-3 (Table 1); this construct was then inserted into pPK2-Bar-GFP [14] to form pPK2-bar-GFP-MrKu70, which was transformed into DMrKu70.

Testing conidial tolerance to abiotic stresses
Tolerance to cold stress (15uC), heat shock (45uC for 2 h), oxidative stress (0.005% H 2 O 2 ) and high osmolarity stress (1.5 M KCl) was tested as previously described [26]. Briefly, tolerance to cold and heat shock stress was shown by the germination rate of conidia (1.5610 6 ) in a Petri dish (diameter = 3cm) containing 2 ml of 0.01% yeast extract (Oxoid Ltd. Hampshire, England). Tolerance to heat shock was investigated by incubating conidia at 45uC for 2 h and then transferring to 26uC for continued growth. Germination was checked every 2 h. Cold stress was induced by incubating the plates at 15uC, and germination was checked at 2-h intervals. Tolerance to oxidative stress and osmolarity stress was tested by measuring the germination rate of conidia at 26uC in yeast extracts supplemented with H 2 O 2 (0.005%) and KCl (1.5 M), respectively.
The tolerance to UV radiation was tested by measuring the germination rate of conidia treated with UV radiation [24]. Briefly, conidia were placed at a final concentration of 5610 5 conidia/mL in 3 ml of 0.01% yeast extract per Petri dish (diameter = 3 cm). Petri dishes were irradiated in a UV-B GS GENE Linker UV Chamber (Bio-Rad, USA) and then either exposed to light energy (photoreactivation) for 4 h by placing the dishes 60 cm from two fluorescent light bulbs (15 W, Sylvania F15T8/CW/SS) or kept in the dark by covering immediately with aluminum foil. Conidia were then incubated at 26uC and germination was checked at 2-h intervals.
Insects were inoculated by immersion in conidial suspensions (1610 7 conidia ml 21 ), and LT 50 values were determined using the SPSS Statistical Package (SPSS Inc., Chicago, IL). All bioassays were repeated three times with 30 insects per replicate.