Random mutagenesis analysis and identification of a novel C2H2-type transcription factor from the nematode-trapping fungus Arthrobotrys oligospora

Arthrobotrys oligospora is a typical nematode-trapping fungus. In this study, 37 transformants of A. oligospora were obtained by REMI (restriction enzyme mediated integration) method and phenotypic properties of nine transformants were analyzed. The nine transformants showed differences in growth, conidiation, trap formation, stress tolerance, and/or pathogenicity among each other and with those of the parental wild-type strain (WT). The insertional sites of the hph cassette were identified in transformants X5 and X13. In X5, the cassette was inserted in the non-coding region between AOL_s00076g273 (76g273) and AOL_s00076g274 (76g274) and the transcription of 76g274, but not 76g273, was enhanced in X5. 76g274p had two conserved domains and was predicted as a nucleoprotein, which we confirmed by its nuclear localization in Saccharomyces cerevisiae using the green fluorescent protein-fused 76g274p. The transcription of 76g274 was stimulated or inhibited by several environmental factors. The sporulation yields of 76g274-deficient mutants were decreased by 70%, and transcription of several sporulation-related genes was severely diminished compared to the WT during the conidiation. In summary, a method for screening mutants was established in A. oligospora and using the method, we identified a novel C2H2-type transcription factor that positively regulates the conidiation of A. oligospora.

oligospora 16 , were characterized. The information from the NTF genomes, proteomes and transcriptomes has contributed to identifying the important genes involved in the regulation of mycelial development, conidiation, trap formation and pathogenicity.
Construction of random mutant library is a useful method for identifying genes of unknown functions in microorganisms. Several methods have been developed and reported in different fungal species, such as restriction enzyme mediated integration (REMI), transposon arrayed gene knockout (TAGKO) and Agrobacterium tumefaciens-mediated transformation (ATMT) 17 . TAGKO has been reported in diverse fungi, such as Fusarium oxysporum 18 and Aspergillus niger 19 . ATMT has been widely applied in genetic transformation of filamentous fungi, such as Magnaporthe grisea 20 , Ustilago maydis 21 and Cordyceps militaris 22 . REMI technique was initially established in Saccharomyces cerevisiae 23 and has been widely used in diverse fungi. In 1999, the hygromycin-B phosphotransferase gene (hph) was transformed to A. oligospora using REMI method, and 7 out of 13 transformants were single copy integrations, whereas the remaining were multiple and more complex integrations 24 . Subsequently, the encoding gene of serine protease PII 6 was transformed into A. oligospora by REMI method, and mutants containing additional copies of the PII gene developed a higher number of infection structures and had an increased speed of capturing and killing nematodes than the wild type strain 9 . Recently, an improved REMI method was applied to transform A. oligospora 25 and Monacrosporium sphaeroides 26 . REMI method can reliably obtain considerable mutant frequencies and single-copy insertion rates compared to TAGKO and ATMT, although the transformation utilizing protoplasts limits its application in fungi 17 .
A. oligospora can capture nematodes by producing three-dimensional networks 2 . The fungus has been a model species for studying the interaction between NTF and nematodes. Recently, the genome of A. oligospora was sequenced; it contains 40.07 Mb assembled sequence with 11,479 predicted genes, and only 44.6% were mapped in the KOG/COG database 14 . Comparative genomic analysis showed that 53.74% genes are specific to A. oligospora 14 . While the majority of these genes had not been annotated, we hypothesize that at least some of them might be involved in the unique biological functions in this fungus, such as trap formation and conidiation. In order to identify novel genes related to trap formation, conidiation and virulence in A. oligospora, 37 random-insertional mutants were screened by REMI method, and nine of them were selected for comparing their physiological and biochemical properties. Furthermore, the insertional sites of hph cassette in the mutants X5 and X13 were identified, and a novel Zinc-finger protein was characterized in A. oligospora.

Results
Random mutagenesis of A. oligospora by REMI technique. The size of the recombinational vector (pMD-hph) is 4.1 kb, including 7 common single restriction sites located in the 5′ and 3′ ends of the hph cassette ( Supplementary Fig. S1), recognized by enzymes XbaI, KpnI, SmaI, XmaI, SacI, SalI and HindIII. The protoplasts of A. oligospora were prepared according to methods described previously 24,27 , and 1 × 10 7 −10 8 protoplasts/mL was used for transformation. The sensitivity of A. oligospora against hygromycin B (Hyg, 200 µg/mL) was used to screen the transformants.
The rate of endonuclease-induced damage to protoplasts was determined using restriction enzymes XbaI (X group) and SmaI (A group), respectively, and 30 U was determined as a suitable concentration for REMI transformation (Supplementary Table S1). In total, 41 (X group) and 7 (A group) transformants were screened and identified (Supplementary Table S1), respectively. Those Hyg-resistant (Hyg r ) transformants were verified by amplifying the hph cassette and five of them were false-positive transformants. Then six Hyg r strains lost the hph cassette after continuous sub-culture on screening medium, thus the genetic stability of the transformants was determined to be 86%, consistent with that in previous reports 25, 26 . Physiological and biochemical analysis of mutants. Nine mutants were randomly chosen to determine their phenotypic and physiological properties. Differences in mycelial growth were observed among the mutants and the wild-type strain (WT) in TYGA and PDA (Fig. 1A). In general, the fungi grew faster in PDA than that in TYGA. Mutants X18 and A2 grew faster than the WT in the TYGA, while X13 grew slower than the WT, and other mutants had no difference to the WT. The growth of X18 was similar to the WT in PDA, while other mutants grew significantly slower than the WT (Fig. 1A). Meanwhile, the conidiation of mutants also showed obvious change compared to the WT, with the conidial production of the strains X5, X28 and A5 increased by 10.8, 5.5 and 2.3 folds (Fig. 1B), respectively. On the contrary, the conidial production of the strains X13, A1 and A2 were decreased by 74%, 62% and 66%, respectively, compared to the WT. Moreover, the conidial germination rate of all mutants was observed to have no significant difference from that of WT. Each conidium of strains X5 and X13 produced one germ tube, same as the WT. However, several mutants (such as A1, A2, X28, A5 and X0) produced two germ tubes from a single conidium, while some mutants had conidia producing either one or two germ tubes, such as A4 and X18 ( Supplementary Fig. S2). After cultured in LMZ for 6 days, the fermentation liquid of the WT showed obvious protease activity, while the proteolytic activity of mutants such as A1, A4 and X0 was decreased, and the other six mutants had no protease activity ( Supplementary Fig. S3).
The WT and eight of the nine mutants could produce traps after induced by nematodes on water agar (WA) plates, while mutant X13 did not produce any trap ( Supplementary Fig. S4). The nematicidal activity of the WT was 9% and 52% when induced by nematodes for 10 and 24 h (Fig. 1C), respectively. However, the nematicidal activities of X5 and A2 were considerably increased, and 33% and 89% nematodes were killed by X5 when induced by nematodes for 10 and 24 h, respectively. On the contrary, several mutants, such as X13, A1 and A5, showed lower nematocidal activity than the WT when induced by nematodes for 24 h.
All mutants grew well on the TG plates, whereas their growths were inhibited when supplemented with NaCl, H 2 O 2 and SDS in medium, similar to that of the WT strain. Four mutants (A2, A5, X0 and X13) could not grow in the TG medium supplemented with 0.05% SDS, while other five mutants (A1, A4, X5, X18 and X28) could grow slightly ( Supplementary Fig. S5). Similarly, strains X5, X18 and X28 could grow on the TG medium supplemented with H 2 O 2 (30 mM), but WT and other mutants could not grow on this medium. Moreover, all mutants could grow on the TG media supplemented with NaCl (0.3 M), but their growth rates were slower than those on the medium without NaCl. A2 grew slower than the WT, and the X18 grew faster than the WT in the TG medium supplemented with 0.3 M NaCl ( Supplementary Fig. S5).
Identification of the insertion sites of the hph cassette in mutants X5 and X13. Flanking sequences of the hph cassette in strains X15 and X13 were amplified by DNA walking method, and the insertion sites were identified by blast analysis in the genome of A. oligospora. The hph cassette was determined to be inserted into the non-coding region (NCR) between AOL_s00078g287 (78g287) and AOL_s00078g288 (78g288) in X13 ( Supplementary Fig. S6), and in the NCR between AOL_s00076g273 (76g273) and AOL_s00076g274 (76g274) in X5 ( Fig. 2A). The insertion sites of the hph cassette in mutants X5 and X13 were further confirmed by PCR amplification using primers X5inserF/X5inserR and X13inserF/X13inserR (Supplementary Table S2), respectively. Specific bands of expected sizes were amplified from the WT and X5 (Fig. 2C), except that a weak 660 bp-band the same size as in the WT was also amplified from X13 ( Supplementary Fig. S6). The 660 bp-band in X13 was sequenced and it was consistent with the sequence of the WT, which suggested that strain X13 lies in a heterozygotic state. Moreover, the insertion of hph cassette in the X5 was further analyzed by Southern blotting using two restriction enzymes, HindIII and SalI (Fig. 2B), respectively, which had only one recognition site in the recombinational vector pMD-hph (Fig. S1). The genomic DNA of the WT and X5 was digested with either HindIII or SalI, respectively, and the hph cassette was used as a probe for Southern analysis. A single and expected size of band hybridizing to the hph cassette was observed in the X5 after its genomic DNA was digested using HindIII or SalI, respectively, while no band was observed in the WT under same condition (Fig. 2D), which suggested that the hph cassette was a single integration in X5.
The transcriptional levels of genes 76g273 and 76g274 were determined by real time PCR (RT-PCR) in WT and X5, respectively. There was about a two-fold increase of gene 76g274 in strain X5 compared to that of WT at 2, 4 and 6 days, respectively, while that of gene 76g273 showed no significant difference compared to that of WT (Fig. 3A). Moreover, the transcriptional levels of genes 76g273 and 76g274 were determined in the WT after induced by nematodes at 0 h and 24 h, respectively. Our results showed that the transcription of gene 76g274 was significantly up-regulated (p < 0.05) after induced by nematodes (Fig. 3B), which suggested that gene 76g274 might be involved in trap formation and infection process in A. oligospora. Analysis and prediction of the protein 76g274p. According to above analysis, gene 76g274 was chosen for further study. The gene 76g274 comprises of an open reading frame containing two introns and three exons. It codes for a polypeptide (76g274p) of 524 amino acid residues without a signal peptide, with a predicted isoelectric point and molecular weight of 9.43 and 57.5 kDa, respectively. Homologous analysis indicated that 76g274p shares a low sequence identity (26.6-35.0%) to analogues from other filamentous fungi, such as Aspergillus fumigatus, Aspergillus niger, Beauveria bassiana, Magnaporthe oryzae and Trichoderma reesei. Interestingly, 76g274p shares a high degree of sequence identity (86.5%) to a homologous protein from Dactylellina haptotyla (syn. M. haptotylum), a nematode-trapping fungus. Phylogenetic analysis also showed that the analogues of 76g274p from different filamentous fungi clustered into two clades (Fig. 4A). Protein 76g274p and its homologous protein from D. haptotyla were clustered in a single clade, suggesting that they share a close evolutionary relationship with each other but distant from other fungi.
The functional domains of 76g274p were analyzed using Interpro 28 , and two conserved domains (Zinc finger C 2 H 2 -type/integrase DNA-binding domain [IPR013087] and Zn(2)-C6 fungal-type DNA-binding domain [IPR001138]) were determined. Alignment analysis of 76g274p and other fungal analogues showed that they share a typical C 2 H 2 -type protein domain (Fig. 4B), however, they share a low identity (26.6-35.0%) on the whole sequences. The amino acid residues in position 145-221 of 76g274p was highly conservative among different fungi, among which the amino acid sequences at positions 155-167 and 171-186 were found to be two typical C 2 H 2 -type Zn 2+ -binding domains. Moreover, sequences in positions 298-324 consisted of the conserved Zn(2)-C6 fungal-type DNA-binding domain (Fig. 4C).
Other conserved features were predicted using PredictProtein 29 and LOCtree 30 . The protein 76g274p was predicted to be located in the nucleus, but no nuclear localization sequence (NLS) similar to known NLS was identified. Additionally, a non-ordinary secondary structure region was predicted in the C-terminal of 76g274p, which was located at positions 395-524. Furthermore, several protein-protein interaction sites were predicted in 76g274p, such as the cAMP-/cGMP-dependent protein kinase phosphorylation site, a protein kinase C phosphorylation site, an N-myristoylation site, a casein kinase II phosphorylation site and a tyrosine kinase phosphorylation site. Analyses via Swiss-Model and STRING showed that a zinc-finger protein structure was located in the conserved domain of 76g274p at positions 151-205 and that it was similar to the zinc-finger domain 2 (no. cltf6D) from Xenopus TFIIIA 31 , which can bind to the 5 s ribosomal RNA gene internal control region. However, little information about the interaction partners of 76g274p was predicted by STRING, suggesting it might be a novel protein with unknown function.
Localization analysis of the protein 76g274p. In order to investigate the expressional location of the protein 76g274p, the cDNA sequence of 76g274 was cloned into the pDD-GFP as pAo76g274-GFP to express GFP-fused 76g274p. pAo76g274-GFP and pDD-GFP were transformed into Saccharomyces cerevisiae (BY4727), respectively. It was observed that only the DAPI-stained nucleus was found in the wild type BY4742 (negative control) (Fig. 5C), and green fluorescent protein (GFP) was uniformly distributed in the cell of strain BY4742::pDD-GFP (Fig. 5B), while GFP signal was centralized in the nucleus of the strain BY4742::pAo76g274-GFP and co-located with DAPI staining (Fig. 5A). Therefore, the gene 76g274 was expressed and located on the nucleus in the model fungus S. cerevisiae.
Expression of gene 76g274 in A. oligospora under different conditions. The transcriptional level of gene 76g274 was determined in A. oligospora incubated on TG medium at 26 °C. Its transcription was shown to constantly increase to a highest level of 33.7 folds for initial four days, and then decreased with the extension of culture times (Fig. 6A). In addition, the transcriptional level of gene 76g274 was also analyzed under various environmental factors (Fig. 6B). For different temperature treatments, the transcription of gene 76g274 decreased slightly when the mycelium was incubated at 12 °C for 1 h, and sharply enhanced at 42 °C for 1 h compared to the 26 °C environment. Among the pH treatments, its transcription was down-regulated at pH 4.0 but up-regulated by 3.87 folds at pH 10.0 for 4 h. For osmotic stress treatments, the transcription of gene 76g274 was down-regulated by treated with 0.5 M NaCl and sorbitol (DS) for 4 h, respectively, with the most inhibition by DS (a decrease of 88%). For other treatments, the transcription of gene 76g274 was up-regulated by 2.2 folds when treated with 0.02% SDS for 4 h, up-regulated by 2 folds treated by vitamin C (1 mg/mL) for 4 h, and up-regulated by 6.6, 4.9 and 2.5 folds by G418, Hyg, and actidione respectively. In contrast, it was inhibited by Congo red (100 µg/mL), ethanol (2%) and DMSO (2%). The transcription of gene 76g274 was, not influenced by treatment with 5 mM H 2 O 2 or carbendazim (300 µg/mL) for 4 h. Among the treatments with metals, the transcription of gene 76g274 was significantly inhibited by Cu 2+ but dramatically increased by Mn 2+ and Fe 2+ .
Disruption of 76g274 impairs conidiation in A. oligospora. The 76g274 gene was disrupted by homologous recombination (Fig. 7A,B) and three transformants (Δ4, Δ6 and Δc) were screened and verified by PCR using primers 274YZ-F and 274YZ-R (Supplementary Table S2) (Fig. 7C), respectively. The growth rate of mutants was faster than the WT strain individually in CMA and WA media, while no obvious difference was observed when growing on PDA medium. Meanwhile, WT and three mutants had no difference in resistance to NaCl (0.3 M), SDS (0.06%) and H 2 O 2 (15 mM). Moreover, difference on the proteolytic activity was not observed between the WT and mutants, and all three mutants produced traps at the same level as the WT. However, the conidiation of mutants was remarkably inhibited and the sporulation yield of mutants was decreased by about 70% (Fig. 7E). In order to identify the potential roles of gene 76g274 in A. oligospora, the mutant Δc was further confirmed by Southern blotting analysis. The single band hybridizing to the 76g274 gene probe was observed in the WT and transformant Δc, and the sizes of the hybridized bands were consistent with expectation (Fig. 7D). The sporulation-related genes in A. oligospora were searched based on the identified homologous genes in the model fungus Aspergillus nidulans, such as AbaA, FlbC, FlbD, FluG, LreA, LreB, MedA, VosA, VeA and VelB (Supplementary Table S3) 32 . The WT and strain Δc were cultured in PL-4 medium 14 at 26 °C at 180 rpm for 3 days, respectively. The mycelia of WT and strain Δc were collected by filtration, and then transferred to sterilized plate (6 cm) for conidia production at 26 °C. The results showed that the conidia began to produce after 52 h, and the conidia number stopped to increase at 124 h. The transcriptional level of sporulation-related genes was determined by RT-PCR at 0 h, 52 h and 124 h, respectively (Fig. 8). The transcriptional pattern of partial sporulation-related genes, such as FluG, VosA, LreA, LreB, VelB, Csp1 and Csp3, was similar between   WT and strain Δc during the conidiation. They transcribed at high level at the early stage of sporulation (0 h) and decreased at middle and later stages (52 h and 124 h, respectively). Interestingly, conidiation-specific gene Csp2 showed differential expressional profile from Csp1 and Csp3. Meanwhile, three septin genes, Sep1, Sep2 and Sep3 also transcribed similarly at different stages of conidiation, and the transcription of abaA on each time point was almost overlapped between WT and strain Δc. Remarkably, transcription of FlbC, FlbD and Med was up-regulated at early and middle stages of sporulation in WT, while they was completely disturbed in strain Δc. In summary, above mentioned results showed 76g274 has positive effects on most tested genes to promote sporulation at different stages.

Discussion
In this study, the sequenced nematode-trapping fungus A. oligospora was used as a model strain, and 37 mutants were obtained using REMI method through protoplast transformation. These mutants showed a high degree of genetic stability. Phenotypic analysis showed that the tested mutants differed from the WT in growth, conidiation, spore germination, trap formation, stress tolerance and nematicidal activity. Most notably, the mutant X5 had an obvious enhancement in conidiation and nematicidal activity, while X13 abolished trap formation and reduced pathogenicity for nematodes. However, the transformation efficiency of the hph cassette was lower than in previous reports 25,26 . In summary, REMI technology was an efficient method for constructing a random mutant library and screening the mutants with phenotypic differences in A. oligospora.
The inserted locations of hph cassette in transformants X5 and X13 were identified using the genome walking method. The hph cassette was inserted into the NCR between genes 78g287 and 78g288 in X13 (Fig. S6), and these two genes were transcribed in opposite directions. The distance between them is 1190 bp, and the location of gene hph was 739 bp from 78g287 and 451 bp to 78g288, respectively, so the hph cassette might be inserted into the promoter region of both genes. Similarly, the hph cassette was inserted into the NCR of genes 76g273 and 76g274 in X5 ( Fig. 2A), and they were also transcribed in opposite directions. The distance between two genes is 8332 bp, and the hph cassette is 264 bp from 76g273 and 8068 bp from 76g274. Moreover, the insertion of pMD-hph was further verified by Southern blotting analysis. We did not observe hybridizing brands with the complete pMD-hph as a probe, likely due to the low efficiency of long probe (>4 kb) in this assay. Fortunately, when we used hph cassette as probe, expected single band was observed in the genomic DNA of X5 digested by either HindIII or SalI, and no band was observed in the WT under same condition. The results of southern blotting with two different restriction enzymes suggest a single insertion event of hph-contained fragment in X5, although it cannot be excluded that the observed phenotypes in the X5 mutant might not be due to the mapped insertion event. Results from PCR and Southern blotting suggested that X5 was a pure mutant with all nuclei containing the hph cassette at the same site. In contrast, some nuclei in X13 had the integrated hph cassette while others did not. The hph cassette might be located in the promoter region of 76g273, but the transcriptional level of 76g273 was not influenced. Unexpectedly, the transcription of 76g274 was enhanced and increased by about two-fold (Fig. 3B), so the insertion of hph cassette might have split a suppressor of 76g274 and broken its negative regulation for 76g274.
Interproscan and other analyses of protein 76g274p showed it contains two conserved domains, Zinc finger C 2 H 2 -type/integrase DNA-binding domain and Zn(2)-C6 fungal-type DNA-binding domain. And a zinc-finger protein structure was predicted in conserved domain of 76g274p using Swiss-Model and STRING, which is high similarity to the zinc-finger domain from Xenopus TFIIIA 31 . Meanwhile 76g274p shared a low homology to other C 2 H 2 -type protein from other filamentous fungi and had a distant relationship with other fungi (Fig. 4A). Moreover, 76g274p was predicted to locate in the nucleus, which was further verified using co-expression technology in S. cerevisiae (Fig. 5). According to above analyses, 76g274p maybe a C 2 H 2 -type transcription factor, and it was located in nucleus in S. cerevisiae, while no conserved NLS was predicted in 76g274p using PredictProtein. We speculated that it may contain an unknown NLS, while the sequence of NLS need to be furtherly identified.
These analyses suggested that 76g274p might be a potential C 2 H 2 -type transcription factor. In fact, C 2 H 2 zinc finger proteins are a major class of transcription factors that play crucial roles in fungal growth, development, various stress responses, and virulence [33][34][35][36][37] . Some C 2 H 2 zinc finger proteins positively regulated fungal  Supplementary Table S3. β-tubulin gene (AOL_s00076g640) of A. oligospora was used as the internal control. *P < 0.05 Δc versus WT. The horizontal bars depict the median, and data are presented as suggested by Weissgerber et al. 58 and Klaus 59 .
conidiogenesis, conidia germination, appressorium formation 33,34 , and the production of aerial mycelium 35 . Most of them were also involved in stress responses, such as responses to high concentrations of Ca 2+ or Na + or high osmolarity. Moreover, many of them were also essential for full virulence 36,37 . However, some C 2 H 2 zinc finger proteins were involved neither in fungal development nor in pathogenicity 38 .
The gene 76g274 was very sensitive to the change of environmental factors. Its transcription was significantly stimulated or inhibited when the fungus A. oligospora was treated with unfavorable conditions (such as extreme temperatures and pH values, high osmotic pressure, and exposure to heavy metal ions, cell wall-damage agents, peroxide, organics and antibiotics) (Fig. 6B). For example, the transcriptional level of gene 76g274 was enhanced remarkably by high temperature, high pH, several metal ions (Mn 2+ , Fe 2+ ) and antibiotics (G418, Hyg and actidione), and its transcription was inhibited significantly by sorbitol, Cu 2+ and Zn 2+ . The sensitivity of this gene in response to variable environments might be beneficial for the fungus to adapt to changing environments, which also helps explain why some fungi show various colonial morphologies in different media 39 .
In order to identify the potential function of 76g274, the deletion mutants were constructed using homologous recombination (Fig. 7A). Phenotypic analysis found that the mutant did not differ significantly from the WT in growth, resistance to NaCl, SDS and H 2 O 2 , proteolytic activity, as well as trap formation, but the mutant had sharply reduced sporulation yields (Fig. 7E). Though the strain X5 seemed to have an increase in pathogencity and stress-resistance compared to 76g274-deficient mutants, it can be explained as the large number of conidia from X5 due to moderate high expression of 76g274 contributed to those phenotypes, because conidia can directly produce traps according previous research and always have stronger anti-stress capacity 10,40 . Meanwhile, the transcription of the majority of sporulation-related genes was significantly down-regulated during the conidiation in mutant Δc (Fig. 8). In strain X5, the transcription of gene 76g274 was increased by about two-fold, and the sporulation yield was increased by 10.8 folds. These results illustrated that 76g274p might be a novel transcription factor, which positively regulated the conidiation in A. oligospora.
There are some shortcomings exist in our present work, such as lacking overexpression and complementation of gene 76g274. In order to confirm furtherly the function of this gene, much work had been done to construct the 76g274-overexpressing strain, while we had not obtained stable-overexpressing strain. The genetic manipulation system to overexpress genes is extremely difficult in this fungus, likely partly due to its multinucleic nature. Meanwhile, we also failed to construct complementation of the 76g274 deletion mutant due to available resistant markers for NTF are rather limited. We had attempted common markers used in other fungi, including glufosinate-ammonium, G418, benomyl, aureobasidin A, phleomycin or pyrithiamine, while they are not effective for NTF, so we could not find the second selective marker to perform complementation analysis at present. Moreover, ATMT had also been attempted to construct the mutant library in A. oligospora in our laboratory, however no satisfactory result was acquired. In recent years, several methods had been developed and used for the genetic transformation in fungi, for example zinc finger nucleases 41 , TALE-nucleases [42][43][44] , Δku70/ku80/Lig4 45 . Especially, the CRISPR/Cas9 sysytem 43,46 had been used in several fungi for successful genetic manipulation and genome edition, such as Trichoderma reesei 47 . NTF play a vital role in control of plant parasitic nematodes, so it is very necessary to develop an efficient method for genetic manipulation in NTF, which would contribute vastly to understanding the molecular mechanism of NTF infection against nematodes and also to screening high-virulent strains for biological control of nematodes.

Materials and Methods
Strains, plasmids and culture conditions. A. oligospora strain ATCC24927 used in this study was purchased from American Type Culture Collection (ATCC) and maintained on potato dextrose agar (PDA) at 26 °C 14 . Saccharomyces cerevisiae (FY834) was used for recombinational cloning procedure and grown in YPD or YPDA plates. SC-Ura plate was used for selecting colonies of yeast strain FY834 harboring correctly recombined constructs 48 . Plasmids (pRS426 and pCSN44) were maintained in Escherichia coli strain DH5а (Takara, Shiga, Japan), respectively. Regeneration medium (PDASS) was used for protoplast regeneration 27 , and PDASS supplemented with 200 µg/mL Hyg (Amresco, Solon, USA) was used for selecting transformants.
S. cerevisiae (FY4742) was used for recombinational cloning procedure and grown in YPD; the plasmid pDD-GFP was used to locate the target gene. Caenorhabditis elegans (strain N2) was grown in oatmeal medium at 26 °C, and the nematode extract was prepared according to our previous report 14 .

REMI transformation
The hph cassette was amplified from the plasmid pCSN44 with primer set Hph-F and Hph-R (Supplementary  Table S2), and the PCR fragment was cloned into vector pMD18-T (Takara, Shiga, Japan), the recombinational plasmid (pMD-hph) ( Supplementary Fig. S1) was isolated using Plasmid Mini Kit (Takara, Shiga, Japan). The pMD-hph was digested using restriction enzyme XbaI or SmaI for further use. Protoplasts of A. oligospora were prepared and transformed according to methods described previously 24,27 . Transformants (mutants) were selected initially on PDASS containing 200 µg/mL Hyg and confirmed further by PCR analysis using primers Hph-1F and Hph-1R (Supplementary Table S2), and the genetic stability of transformants was detected according to methods described previously 24, 27 . Analysis of resistance of A. oligospora and transformants to stress conditions. The TG medium was used as basic medium, NaCl (0.3 M), H 2 O 2 (30 mM) and SDS (0.05%) were supplemented in TG, respectively. The WT and mutants were incubated respectively on above medium at 26 °C for 7 d, and the growth rate was determined and colonial morphology was observed.
Vegetative growth and conidia yield comparisons. The mycelial plugs of the WT and mutants of the same size and age were incubated on PDA and TYGA media at 26 °C for 7 days, respectively, the growth rate and morphology of colony were observed and determined. Sporulation capacity of WT and mutants was determined according to our previous report 27 . Each experiment was performed with three biological replicates.
Protease activity assays. The WT and mutants were incubated on LMZ medium 49 at 26 °C, respectively.
The fermentation liquid was collected after culturing for 6 days. Protease activity was assayed on casein-plate using the method described by previous report 49 .

Induction of trap formation and virulence assays. Trap formation in A. oligospora induced by nem-
atodes and nematode extract according to our previous report 27 . The plate was incubated at 26 °C for 3-4 days, 80-100 adult nematodes were added to the plate. After 12 h, mycelia and traps were observed at specified time intervals under light microscopy (Olympus, Tokyo, Japan). Meanwhile, the number of captured nematodes was counted using a light microscope 27 . The bioassay was performed with three biological replicates.
Identification of the insertion sites of the hph cassette in strains X5 and X13. To identify the insertion site of the hph cassette from strains X5 and X13, two nested PCR primers (Supplementary Table 2) were designed according to the nucleotide sequence of the hph cassette. The 5′ terminal fragment of the hph cassette was amplified using the nested PCR primers and the DNA-Walking Speedup ™ Premix Kit (Seegene, Korea), respectively. PCR amplification was performed according to the user's manual. The PCR product was cloned and sequenced, and the insertion site was identified by blast analysis in the genome of A. oligospora. Meanwhile, the mutation site was further identified by PCR amplification using two pairs of primers in X5 and X13, respectively (Supplementary Table 2). Southern analysis was carried out according to the instructions provided by the North2South ® chemiluminescent hybridization and detection kit (Pierce, Rockford, USA). The probe was amplified using pair X5-S1/X5-S2 (Supplementary Table S2), and restriction enzymes SalI and HindIII were used to digest the genomic DNA for Southern analysis.
Sequence and phylogenetic analysis of 76g274p. The sequence of 76g274p (EGX50199) in A. oligospora was downloaded from the GenBank. The theoretical isoelectric point (pI) and molecular weight of 76g274p were calculated using the pI/MW tool 50 , and conserved functional domains were analyzed via the InterPro Web site 28 . Moreover, the structure of 76g274p was predicted by PredictProtein 51 , and its tertiary structure was analyzed by Swiss-Model 52 and Phyre2 53 . Meanwhile, the protein-protein interaction was predicted using STRING 9.05 54 .
The amino acid sequence of 76g274p was retrieved and its analogues in different fungi were downloaded from GenBank. A neighbor-joining tree was constructed using the Mega 5.1 software package 55 . Quantitative PCR (RT-PCR) analysis. Primers were designed and synthesized based on the DNA sequences of target genes, including 76g273 and AOL_s00076g274 (Supplementary Table S2) and primers of sporulation-related genes in A. oligospora were also designed using Primer3 (Supplementary Table S3). Total RNA was extracted with Trizol Reagent (Invitrogen, Carlsbad, California), purified with a RNeasy Mini Kit (Qiagen, Valencia, Califonia), and then reverse transcribed with PrimeScript RT reagent Kit (Takara, Shiga, Japan) by following the manufacturer's instructions. RT-PCR was conducted according to our recent report 56 . β-tubulin gene (AOL_s00076g640) of A. oligospora was used as the internal control. RT-PCR was performed with three biological replicates for each gene.
Locolization analysis of the gene 76g274 in S. cerevisiae. The cDNA sequence of 76g274 without terminator (TAA) was amplified using primer pair 76g274F/76g274R, and it was inserted into the linearized vector pDD-GFP (digested by SmaI), the final vector pAoG274-GFP was transformed into S. cerevisiae (BY4742), and tranformants were screened on SC-Ura plate. The strain BY4742, BY4742::pDD-GFP and BY4742::pAoG274-GFP were cultured in YPD at 30 °C for 12 h, three strains were collected, respectively, and observed under fluorescence microscope.
Deletion of the gene 76g274. The gene 76g274 replacement construct was generated using a modified yeast cloning procedure 48,57 . Two fragments corresponding to the 5′ and 3′ of the gene 76g274 were amplified with primer sets 274-5 F/274-5 R and 274-3 F/274-3 R (Supplementary Table S2), respectively. The two fragments were co-transformed into yeast strain FY834 along with the hph cassette and gapped yeast shuttle vector (pRS426). The final disruption vector (pRS426-76g274-hph) was recovered by transformation into E. coli DH5a. Protoplast of A. oligospora was prepared and transformed according to our recent report 27 . Transformants were selected on PDASS containing 200 µg/mL Hyg, and further confirmed by PCR and Southern blot analyses. Southern analysis was carried out according to above description, and the probe was amplified using primer pair 274-S1/274-S2 (Supplementary Table S2), and restriction enzyme BamHI was used to digest the genomic DNA.
Statistical analysis. Medians were calculated as recommended for biomedical research 58,59 . One-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test is indicated when used, (P < 0.05). All statistical analyzes were made using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, California, USA).