The wild-type strain HWD2 and all the transformants of U. virens generated in this study were routinely cultured on potato sucrose agar (PSA, 2% sucrose plus extract from boiled peeled potato) at 28°C, and stored in the form of mycelial-colonized filter paper at -20°C. The A. tumefaciens strain EHA105 and binary vector pTFCM were used for U. virens transformation. Plasmids KS1004 and pneoP3300III were used for gene disruption or complementation vector construction.

The susceptible rice cultivar Wanxian 98 was used in virulence assays. The seeds were kept for 24 h at 30°C before planting. After 10 days, four seedlings were placed into pots (25 cm × 20 cm × 30 cm, length × width × height) each containing 5 kg of autoclaved paddy soil. In the greenhouse, pots were fertilized twice (4 g carbamide per bucket): once at tillering (after 45 days of growth) and just before inoculation at the at the booting stage (after 90 days of growth; Jia et al., 2015).

Agrobacterium-mediated transformation was carried out following the protocols described Yu et al. (2015) with minor modifications. The wild-type strain HWD2 was cultured in a 250 mL flask containing 150 mL liquid potato sucrose broth (PSB). The flask was placed in a shaking incubator at 28°C in the dark. After shaking at 160 rpm for 7 days, the cultures were filtered through multiple layers of cheese cloth, and conidia were obtained from the filtrate by centrifugation (3,000 rpm for 5 min). The conidial suspension was adjusted to 1 × 10⁶ conidia per mL using a haemocytometer.

The A. tumefaciens strain EH105 was grown at 28°C with shaking at 180 rpm for 48 h in minimal medium supplemented with kanamycin (50 μg/mL). Then, A. tumefaciens cells were grown in induction medium supplemented with 200 μM acetosyringone. After shaking at 180 rpm for an additional 10 h at 28°C, bacterial cultures were diluted to an optical density of 0.5 OD units at 600 nm and were mixed 1:1 with a conidial suspension from HWD2 (10⁶ spores/mL). The mix was plated onto co-cultivation medium with a layer of nitrocellulose filter. After co-cultivation at 24°C for 4 days, the membrane was removed, and placed mycelium-side down onto PSA containing 500 μg/mL of cefotaxime to counter-select bacteria, and 200 μg/mL of hygromycin to select for U. virens transformants. After incubation at 28°C for 5–7 days, transformant colonies were transferred to PSA plates containing 200 μg/mL of hygromycin for a second round of selection.

To test for the mitotic stability of the integrated hygromycin resistance cassette, 20 randomly chosen transformants were cultivated on PSA without hygromycin. After weekly transfer to new plates for four passages by subculturing of hyphal tips, transformants were grown on PSA plates containing hygromycin (200 mg/mL).

The fungus was propagated in PSA plates for 14 days at 28°C. Then, 3-mm-diameter mycelia dishes were cut from the edge of a colony and inoculated in a 50 mL flask containing 30 mL PSB which was placed in a shaking incubator. After shaking at 180 rpm for 7 days, the cultures were filtered through three layers of gauze, and conidial production was measured using a haemocytometer. The experiment was repeated three times with three replicates each time.

Genomic DNA sequences of the transformants flanking T-DNA insertions were amplified by TAIL-PCR (thermal asymmetric interlaced-polymerase chain reaction) and inverse PCR with primer sequences shown in Supplementary Table 1. For TAIL-PCR, genomic DNA was used as a template in successive reactions with nested left border primers (LB1, 2, and 3) and right border primers (RB1, 2, and 3) together with the degenerate primers (AD1, 2, 3, or 4). PCR settings for TAIL-PCR followed Liu et al. (2013). For inverse PCR, genomic DNA was digested with SacI and circularized with T4 DNA ligase (Invitrogen, Karlsruhe, Germany). The product was purified using a Nucleic Acid Purification kit (Axygen, Union City, CA, USA). The reaction conditions for first round PCR were: 1 cycle at 95°C for 5 min, 30 cycles of 95°C for 30 s, 55°C for 45 s, and 72°C for 4 min and a final cycle at 72°C for 5 min. The second round nested PCR was performed with the same PCR program using 1 ml of the first round PCR product (diluted 1: 50) as a template together with nested primers (Liu et al., 2013). Flanking sequences recovered by TAIL-PCR and inverse PCR were analyzed with the BLAST tool hosted by the National Center for Biotechnology Information¹ against the GenBank database and the genome sequences of U. virens (NCBI, JHTR00000000.1). Nucleotide sequences were compared with known protein sequences using BLASTX (NCBI²). Open reading frames (ORFs) were analyzed using FGENESH (Softberry Inc., Mount Kisco, NY, USA), conserved domains were detected by comparison to the Conserved Domain Database of NCBI³.

The full sequence of UvPRO1 was obtained from the genome sequence of U. virens (NCBI, JHTR00000000.1). To confirm sequence presence, the primers UvPRO1F and UvPRO1R (Supplementary Table 1) were designed and used for the amplification of the UvPRO1 gene from HWD2 isolates. Primers were all designed using Primer Premier 5.0⁴, and ORFs were analyzed using FGENESH. Protein domain and motif predictions were performed with SMART software⁵.

The PRO1 protein sequences from different organisms were obtained from the GenBank database, using the BLAST algorithm with the UvPRO1 sequence. Sequence alignments were performed using the Clustal X (version 2.0⁶), and a phylogenetic tree was generated with Mega software (version 7.0⁷) using the Neighbor-Joining method.

To assess the function of UvPRO1, which was potentially mutated in T133, a vector was constructed for the targeted disruption of UvPRO1 by means of homologous recombination. Vector KS1004 was constructed by cloning a 1.9 kb PtrpC-hph cassette into the SmaI site of pBluescriptII KS, and the hygromycin resistance was used as the first selectable marker for screening of disruption transformants. Vector pneoP3300III was generated by cloning a 2.1 kb neomycin resistant gene cassette into the XbaI site of pCAMBIA3300, and the neomycin resistance was used as the second selectable marker.

A pair of gene-specific primers, UvPRO1F1F and UvPRO1F1R (Supplementary Table 1), was used to amplify the 900 bp fragment (Figure 4A) in the 5′ coding region of UvPRO1. Another pair of gene-specific primers, UvPRO1F2F and UvPRO1F2R (Supplementary Table 1), was used to amplify the 978 bp fragment, containing part of the 3′ coding region of UvPRO1 (Figure 4A). The 900 bp HindIII/SalI-fragment (5′ region of UvPRO1) and the 978 bp XbaI/KpnI-fragment (3′ region of UvPRO1) were cloned into the corresponding restriction sites of the vector KS1004, resulting in the preliminary vector KS1004-UvPRO1. The hph-UvPRO1 cassette (with a 900 bp HindIII/SalI-fragment, a 1909 bp hph-fragment, and a 978 bp XbaI/KpnI-fragment) was cloned into pneoP3300III, resulting in the gene disruption vector p3300neoUvPRO1.

This vector, p3300neoUvPRO1, was transformed into A. tumefaciens EHA105 by electroporation, and then hyphae were transformed with the ATMT protocol. To find UvPRO1 disruption transformants, cultures were grown on PSA amended with hygromycin (200 mg/mL), and then subcultured onto PSA amended with 800 μg/mL of antibiotic G418 (Amresco, Solon, OH, USA). Gene disruption transformants were subjected to PCR with two pairs of primers, UvPRO1KF/UvPRO1KR and HphF/HphR (Supplementary Table 1), and amplicons were detected by PCR and Southern blot analysis.

To confirm targeted gene disruption, the disruption mutant ΔUvPRO1-27 was complemented with a full length sequence of UvPRO1. Because UvPRO1 disruption mutants were unable to grow on the PSA supplemented with G418, the neomycin resistance cassette was chosen as a selectable marker for the complementation transformation. The complementation plasmid p3300neoUvPRO1-Com was based on pneoP3300III. The 3,315 bp UvPRO1 fragment (UvPRO1 ORF plus 574 bp 5′-flanking and 905 bp 3′-flanking sequences) was amplified from genomic DNA of the wild-type with the primer pair UvPRO1ComF and UvPRO1ComR (Supplementary Table 1), and cloned into the BamHI site of pneoP3300III to generate the complementation plasmid p3300neoUvPRO1-Com. To obtain the UvPRO1 complementation transformants, ΔUvPRO1-27 was transformed with vector p3300neoUvPRO1-Com by the ATMT method. The complementation transformants were screened on PSA containing 800 μg/mL G418, and gene fragments were detected by RT-PCR.

Genomic DNA was extracted using CTAB (Sambrook et al., 1989). For Southern blot analysis of T-DNA insertion in U. virens, PCR was used to confirm the presence of T-DNA insertions by using primers HphF and HphR (Supplementary Table 1) to amplify an 887 bp internal region of the hygromycin resistance gene (hph). DNA from the wild-type and the transformants was completely digested with SacI, which has only one recognition site in the binary vector pTFCM, and then size-fractionated through a 0.8% agarose gel and mounted onto a positively charged nylon membrane (Figure 2A). The hph gene was excised from the pTFCM vector and labeled with digoxigenin (DIG)-dUTP using the PCR DIG probe synthesis kit (Roche, Mannheim, Germany) following manufacturer’s instructions. Hybridization was detected using a DIG luminescence detection kit. For Southern blot analysis of UvPRO1 disruption mutants, genomic DNA from the wild-type and the putative UvPRO1 disruption mutants were digested with SacI at 37°C for 24 h. The nylon membrane was hybridized with probe P (Figure 4A).

Hyphae harvested from PSB medium were collected at different points in time (3, 4, 5, 6, 7, 8, and 9 days), as well as inoculated spikelets at different points in time (1, 2, 3, 4, 6, 8, 10, and 12 days). These were frozen in liquid nitrogen and stored at 80°C until required. RNA was extracted using a TRIzol Plus RNA purification kit (Invitrogen, Carlsbad, CA, USA). DNA contamination was removed by DNaseI treatment (RNase free; TaKaRa, Dalian, China). First-strand cDNA was synthesized by using a RevertAid^TM first strand cDNA synthesis kit (Fermentas, St. Leon-Rot, Germany). Expression of UvPRO1 in disruption mutants and the complementation strain were examined by RT-PCR, and a 1560-bp fragment was amplified with gene-specific primers qRT-UvPRO1F and qRT-UvPRO1R (Supplementary Table 1). PCR conditions used 25 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min, with a final extension at 72°C for 5 min.

Expression of UvPRO1 at different developmental stages of the fungus in vitro or in planta was analyzed by qRT-PCR with UvPRO1 gene-specific primers qRT-PRO1F/qRT-PRO1R (Supplementary Table 1). PCR conditions were 40 cycles of 94°C for 15 s, 55°C for 20 s, and 72°C for 15 s, and with a final extension from 65°C to 95°C (0.5°C/5 s; Gu et al., 2012). The U. virensα-tubulin2, as the reference gene, was amplified with primers α-tubulin2F and α-tubulin2R (Supplementary Table 1). PCR reactions were run on a PTC-200 DNA Engine Peltier thermal cycler (BioRad, Hercules, CA, USA). The whole experiment was repeated three times.

For mycelial growth, mycelial plugs (5 mm in diameter) were transferred from 12-day-old PSA plates and grown on fresh PSA medium at 28°C. After 6 and 12 days of being cultured, the radial growth of vegetative mycelia was measured. For conidial production, strains were grown in PSB medium at 28°C. After shaking at 180 rpm for different lengths of time (4, 5, 6, 7, 8, and 9 days), the cultures were filtered through three layers of gauze, and conidial production was measured using the haemocytometer. For testing the sensitivity to various stress chemicals, the strains exposed to CM medium containing either exogenous 0.1–0.5 M NaCl, 0.01–0.05% SDS, or 30–70 mg/L CR were assessed also by measuring colony diameter of 14-day cultures. Each treatment was repeated three times.

For pathogenicity analysis, mycelial plugs of the wild-type, UvPRO1 knock out and complementation strains were transferred from 12-day-old PSA plates and grown in PSB medium at 28°C. After shaking at 180 rpm for 5 days, the cultures were homogenized in a blender, and rice plants were inoculated with 2 mL of mycelial suspension using a syringe in the middle section of distal internodes at the eight stage of panicle development. The rice plants were placed in a plant growth chamber (Wuhan Ruihua Instrument and Equipment Co., Ltd., Wuhan, China) equipped with a high pressure sodium lamp (12 h light/dark cycle) with conditions set at a RH of 95 ± 5% and a temperature of 25 ± 1°C. After a post-inoculation surface wetness period of 120 h, plants were transferred to a greenhouse equipped with an automatic climate control system set at 28 ± 2°C and 75 ± 7% RH. This experiment was repeated three times (Jia et al., 2015). Five of the injected panicles were sampled at each time point (1, 2, 3, 4, 6, 8, 10, and 12 days after inoculation) and others were used to count the severity of false smut infection 15 days after inoculation.

The samples for scanning electron microscopy were first fixed with 2.5% (v/v) glutaraldehyde in 50 mM phosphate buffer (pH 7.2) for 6–8 h at 4°C, before a rinse with the same buffer for 2 h. They were then fixed in 1% (w/v) osmium tetroxide in 50 mM phosphate buffer for 1 h. After dehydration in a graded acetone series, the samples were critical-point dried, mounted on stubs, sputter coated with gold-palladium, and viewed using a JEOL JSM-6390LV scanning electron microscope operating at 10 kV (Hu et al., 2014).

The quantitative data were analyzed with DPS statistical analysis software (version 3.01, China Agric. Press, Beijing, China), using ANOVA. When significant treatment effects were found (P < 0.05), separation of means was done using Fisher’s Least Significant Difference test.

Using the modified protocol for ATMT, a total of 3,016 hygromycin-resistant transformants of U. virens were obtained. The mitotic stability of the integrated T-DNA was tested by analysis of 20 randomly selected transformants. Transformants were serially subcultured for five time on PSA medium not containing hygromycin. Transformants retained the integrated T-DNA, as indicated the ability to grow on PSA containing hygromycin.

All of the 3,016 transformants were screened for sporulation deficiency and five transformants with sporulation deficiency were found. Four transformants (T420, T896, T1296, T2328) were found to have significantly (P < 0.05) lower conidial production, and one transformant (T133) was found to have no conidia (Figure 1). Southern blot analysis showed that, among several mutants with sporulation deficiency, four (T133, T420, T896, T2328) contained a single T-DNA insertion and T1296 contained two T-DNA copies (Figure 2B). T-DNA flanking sequences were recovered from these mutants by amplifying genomic DNA sequences flanking T-DNA insertions of transformants with TAIL-PCR and inverse PCR (Supplementary Table 1). These sequences were used to screen the GenBank database and the genome sequences of U. virens (NCBI, JHTR00000000.1). FGENESH was used to identify ORFs around the T-DNA insertion site. ORF sequences were compared against protein sequences from NCBI with BLASTX. Details on affected genes and disruption sites are shown in Table 1.

In mutant T420, the targeted gene encodes a hypothetical protein, showing similarity to a protein of unknown function from Ophiocordyceps unilateralis (GenBank KOM18477.1). In mutant T896, a single insertion was located inside a predicted ORF of a gene with significant similarity to the Ser/Thr protein phosphatase gene of Metarhizium acridum (GenBank XP_007807767.1). In mutant T1296, one insertion was located upstream of a gene that showed high sequence similarity to a hypothetical protein from Acinetobacter schindleri (GenBank WP_004893973.1), and the other failed during cloning. In mutant T2328, T-DNA targeted upstream of a gene with significant similarity to a polysaccharide synthase Cps1 gene of Fusarium fujikuroi (GenBank KLO79232.1). In mutant T133, T-DNA was inserted into a predicted ORF of a PRO1 gene encoding C6 transcription factor that showed high sequence similarity to a PRO1 gene of M. acridum (GenBank XP_007812494.1; Table 1). The mutant T133 was characterized by no sporulation and contained a single T-DNA insertion. Therefore, our subsequent work focused on the gene UvPRO1 in the mutant.

The aligned sequences of overlapping DNA fragments of the PRO1 gene amplified by PCR from U. virens genomic DNA and from corresponding mRNA revealed a 2,440-bp ORF. The coding domain was predicted to encode a polypeptide consisting of 611 amino acids and a high level of sequence identity (84%) with transcriptional regulatory protein PRO1 of M. acridum CQMa 102. Sequence analysis with SMART revealed that UvPRO1 contained a Fungal_trans_2 conserved domain (Figure 3A).

Phylogenetic analysis of UvPRO1 (GenBank KDB14867.1) to other PRO1 proteins (Figure 3B) revealed that UvPRO1 was most similar to PRO1 proteins of Pochonia chlamydosporia and species of Metarhizium (with identities above 81%), and more distant from those of other fungi (with identities above 58%). This result indicates that PRO1 proteins are conserved among fungi tested.

A gene disruption vector, p3300neoUvPRO1, containing the hph gene and both the 3′ and 5′ flanking regions of UvPRO1, was constructed with two vectors, KS1004 and pneoP3300III (Figure 4A). Vector p3300neoUvPRO1 was transformed into the wild-type, and transformants were selected on hygromycin-containing medium and on G418-containing medium. Among 628 hygromycin-resistant transformants, three without resistance to G418 were obtained.

Only one candidate disruption transformant ΔUvPRO1-27 was found lacking the 1208-bp UvPRO1 fragment compared to the wild-type strain after PCR amplification with PRO1F/PRO1R (Supplementary Table 1); however, an 887-bp hph fragment was obtained by PCR amplification with hphF/hphR (Supplementary Table 1) in this candidate transformant. Furthermore, Southern blot analysis showed that single integration events had occurred in selected UvPRO1 knockout transformant ΔUvPRO1-27 (Figure 4C), which had the 6.3-kb SacI fragment, while the wild-type strain HWD2 had the 1.5-kb SacI fragment. Null mutation of the UvPRO1 gene was further confirmed by RT-PCR analysis, since the UvPRO1 transcript was not detected in the targeted disruption transformant. These results demonstrated that the UvPRO1 gene was deleted in the UvPRO1 disruption transformant ΔUvPRO1-27. To investigate whether altered growth phenotypes and the loss of virulence in UvPRO1 disruption transformants could be restored by reintroduction of a wild-type copy of UvPRO1, we transformed ΔUvPRO1-27 with plasmid pNeo3300IIIUvPRO1-Com. Subsequently, complementation transformant CΔUvPRO1-27 was confirmed by RT-PCR analysis (Figure 4B) and was selected for further studies.

The morphology of the strains was monitored on PSA medium. The ΔUvPRO1-27 mutant produced white colonies with long and abundant aerial hyphae, in contrast with the colonies with a light yellow center surrounded by a white edge of the wild-type rescued strain CΔUvPRO1-27 (Figure 5A). Furthermore, the ΔUvPRO1-27 strain had a reduced apical extension rate (2.6 mm/d), producing smaller colonies than the wild-type (2.8 mm/d) and CΔUv-PRO1-27 (2.8 mm/d), and the mycelial growth rate measured at 6 days on PSA of the ΔUvPRO1-27 strain (2.20 mm/d) was significantly less than the wild-type (2.73 mm/d; Figure 5B).

In PSB medium, mycelia from wild-type strains and the rescued strain CΔUvPRO1-27 produced hyphae with conidiophores at their tips after 5–6 days, and conidia were produced after 7 days at 6.7 or 6.8 × 10⁶ conidia/mL, respectively. However, mycelia of the ΔUv-PRO1-27 produced hyphae without conidiophore formation at 6 days, and no conidia were observed up to 9 days (Figure 5C).

Because the mycelial growth of the UvPRO1 mutant was interrupted, we further monitored the effects of hyperosmotic and cell membrane stresses on CM medium with 0.1–0.5 M NaCl, 0.01–0.05% SDS, or 30–70 ug/mL CR. In the presence of 0.1–0.5 M NaCl, the growth rate of all strains decreased, and ΔUvPRO1-27 mutant displayed more sensitivity under salt stress compared to the wild-type and CΔUvPRO1-27, and the growth rate of ΔUvPRO1-27 mutant was reduced by 16–65%, respectively (Figure 6A). These results suggested that the UvPRO1 may play a role in regulation response to hyperosmotic conditions in U. virens.

We also assayed the effects of SDS and CR treatments that mimic cytoplasm membrane and cell wall stresses, respectively. On CM with 0.01, 0.03, or 0.05% SDS, the growth rate of the ΔUvPRO1-27 mutant was, respectively, reduced by 37.5, 74.6, and 86.4%, while the decrease was 22.1, 31.5, and 61.3% in the wild-type and 23.2, 32.2, and 59.8% in CΔUvPRO1-27 (Figure 6B). In the presence of 30–70 mg/L CR, similar results to growth assays with SDS were obtained, in that ΔUvPRO1-27 mutant displayed a slower radial growth rate than the wild-type or CΔUvPRO1-27 (Figure 6C). These results suggested that the UvPRO1 mutant also had increased sensitivity to CR and SDS. Therefore, UvPRO1 may be involved in regulating responses to membrane and cell wall stresses in U. virens.

Pathogenicity assays of the wild-type strain, ΔUvPRO1 mutant, and UvPRO1 complementary strain were performed on a susceptible host (Wanxian 98). Since the ΔUvPRO1 mutant produced no conidia, we also used a mycelial suspension of the wild-type strain and a UvPRO1 complementary strain for inoculation by injection as well as ΔUvPRO1 mutant. The inoculated plants were examined for colonization and infection by U. virens until 12 dpi. At 1–3 dpi, for the wild-type strain HWD2 and UvPRO1 complementary strain, many hyphal strands were observed to be elongated and extended along the surface of the spikelets (Figures 7D–F). At 4–6 dpi, hyphae were observed on the inner surfaces of spikelets, and filaments were infected by masses of hyphae (Figure 7G). At 7–8 dpi, the florets were covered profusely by hyphal growth with some wrapped around stamens and pistils (Figures 7H,I). At 9–12 dpi, the spaces in the spikelets were filled up by white mycelia (Figure 8A), and large mycelial masses grew out of the spikelets forming smut balls. After 15 dpi, 88.6% of the wild-type strain inoculated plants developed typical symptoms of false smut (Figure 8B), and similarly 86.2% in the UvPRO1 complementary strain (Table 2). For the ΔUvPRO1 mutant, at 1–3 dpi, hyphae were observed to be elongated and extended along the surface of spikelets (Figure 7A), which was similar to that of the wild-type strain and UvPRO1 complementary strain. At 4–6 dpi, mycelia became dehydrated and failed to grow further on spikelets (Figure 7B). No hyphae were observed inside of spikelets until 15 dpi (Figures 7C and 8A,B). This indicated that the ΔUvPRO1 mutant lost the ability for invasive growth on spikelets, and that the UvPRO1 is important for the pathogenicity of U. virens. Therefore, we conclude that the UvPRO1 plays an important role in virulence of U. virens.

We first evaluated the UvPRO1 expression levels of U. virens in PSB medium using qRT-PCR. The results showed that lower expression levels were detected during early vegetative growth stages between 1 and 5 days, and were significantly increased during the conidiation stage between 6 and 9 days (Figure 9A). UvPRO1 expression during spikelet infection stage (4–12 dpi) was much higher than that in the early developmental stages (1–3 dpi), while the relative expression levels of UvPRO1 at 8 dpi was more than twofold higher than that at 1–2 dpi (Figure 9B).