The C. magnum wild-type (WT) strain CAASZK4 was isolated from watermelons in Kaifeng City, Henan Province, China (Guo et al., 2022). The strain AGL-1 of A tumefaciens, containing plasmid pATMT1 which carries a hygromycin B resistance cassette, was used as a T-DNA donor for fungal transformation (Zheng et al., 2011; Li et al., 2019).

The WT strain (CAASZK4) were grown on potato dextrose agar (PDA) medium complemented with a range of hygromycin B concentrations (0, 20, 40, 60, 80, 100, 150, 200, and 250 mg/L) and were incubated at 27°C in darkness for 7 d.

ATMT-based transformation of C. magnum was performed according to the protocol described by Cai et al. with slight modifications (Cai et al., 2013). The AGL-1 strain grew at 28°C in 5 mL of Luria–Bertani (LB) medium supplemented 25 mg/L rifampicin and 50 mg/L kanamycin. Cultures were centrifuged again. Finally, the cells were resuspended in induction medium (IM, Bundock et al., 1995) supplemented with acetosyringone (AS, 200 μM), and incubated for 6 h to achieve OD600 nm values (0.6–0.8).

C. magnum grew on synthetic nutrient-poor agar medium (SNA, Nirenberg, 1976) for 25 d at 27°C in the dark. Subsequently, the conidia of C. magnum were cleaned thrice sterile water and diluted in IM to achieve different concentrations (1 × 10⁶–1 × 10⁸ spores/ml). Aliquots of 100 μL of bacterial culture and 100 μL of conidial suspension were plated onto the 0.45 μm pore nitrocellulose filter of the co-cultivation medium (CM) containing 200 μM AS. After cocultivation for 84 h at 23°C in the dark, the membranes were transferred to PDA containing 200 mg/L cefotaxime, 100 mg/L timentin, and 80 mg/L hygromycin B, and cultured for 18 d at 27°C in the dark. Mycelial plugs from the edges where fresh mycelia were cultured on PDA containing 80 mg/L hygromycin B were used to confirm resistance. To determine the mitotic stability of the transformants, 40 randomly selected transformants were cultured on PDA without hygromycin B for four generations and then transferred back to PDA with hygromycin B.

The transformants and the WT strains of C. magnum were transferred to 100 mL fresh liquid PDA and cultured for 20 d at 27°C in the dark. Genomic DNA of 40 transformants and the WT strains of C. magnum were extracted by CTAB (Sangon Biotech, China).

To analyze genome integration of the transferred genes, the hph gene was amplified using the primer pair hph-F + hph-R (Table 1) and sequenced. PCR was were performed according to the protocol described by Guo et al. with slight modifications (Guo et al., 2022). The annealing temperature of hph was 60°C. PCR amplicons were purified and sequenced by Sangon Biotech Company (Shanghai, China).

The digestion of genomic DNA from 40 transformants or WT strains with EcoRI and HindIII (NEB, Ipswich, United States), electrophoresised on 0.75% agarose, were depurated, and transferred to nylon membranes (Hybond-N+; General Electric Company, Boston, United States). In the hybridization experiments, the digoxigenin (DIG) labeled probe corresponded to 1,380 bp hph of T-DNA from plasmid pATMT1 (Figure 1), and the hph gene was amplified using the primers hph-F and hph-R. Hybridization and chemiluminescence detection of the hybridized Dig-labeled probes were performed using the DIG High Prime DNA Labeling and Detection Starter Kit (Roche, Germany).

Mycelial plugs (5 mm) from 1,460 transformants and WT strains of C. magnum were cultured in PDA at 27°C in the dark for 5–15 d. The pathogenicity of these transformants and WT strain was evaluated on healthy detached watermelon leaves (Citrullus lanatus cv. Hongheping, 5–6 true leaves), which were previously washed with tap water, surface-sterilized with 75% ethanol for 30 s, and rinsed thrice with sterile water. Mycelial discs (5 mm in diameter) were individually placed on the right half of the leaves. PDA blocks (5 mm in diameter) were placed on the left half of the same leaves as controls. All leaves were incubated in sealed sterile plastic plates at 27°C in the dark. Leaf symptoms were assessed at 4 d post-inoculation. All transformants and WT strains were tested in triplicates.

Mycelial plugs (5 mm) from transformants and WT strains were transferred on 60 mm Petri dishes containing SNA medium, and incubated in the dark at 27°C for 25 d. The mycelia, conidia, and appressoria were observed under a light microscopy (Olympus BX51, Japan). The conidia formed in 60-mm Petri dishes were cleaned thrice with sterile water and resuspended in 1 mL sterile water. The number of conidia was calculated using a hemocytometer. To induce the formation of appressoria, the conidia were suspended in 1% glucose and incubated at 27°C in the dark. After 48 h, the germination rate of the conidia was measured. All transformants and WT strains were tested in triplicates.

The T-DNA flanking sequences inserted into the genome were cloned using thermal asymmetric interlaced PCR (TAIL-PCR) protocol with minor modification (Liu et al., 1995; Mullins et al., 2001). Right border primers (R1, R2, and R3) and left border primers (L1, L2, and L3) and arbitrary degenerate primers (AD1, AD2, and AD-3) were utilized for TAIL-PCR (Table 1). PCR conditions were set as described by Mullins and Sessions (Mullins et al., 2001; Sessions et al., 2002), with minor modifications (Supplementary Table S1). The third TAIL-PCR products of all transformants displaying the highest brightness were purified using the FastPure Gel DNA Extraction Mini Kit (Vazyme, China). This fragment was ligated to the vector pTOPO-Blunt, which was transferred into Escherichia coli strain Top10 using a CV16-Zero Background pTOPO-Blunt Cloning Kit (Aidlab, China). E. coli harboring the pTOPO-Blunt vector was sequenced by Sangon Biotech Company (Shanghai, China). To isolate the tagged genes, the sequences flanking the T-DNA of each transformant were used to search the local genome database of WT C. magnum by BLAST+.¹ Gene structure was predicted using the C. fioriniae, C. gloeosporioides, C. graminicola, C. higginsianum, C. orbiculare, and C. sublineola by the FGENESH program.²

Four micrograms of high-quality genomic DNA from three transformants (Cm699, Cm854, and Cm1078) were extracted from fresh mycelia using CTAB (Sangon Biotech, China) and used to construct a sequencing library, following the manufacturer’s instructions (Illumina Inc.). Paired-end sequencing libraries with an insert size of approximately 200 or 400 bp were sequenced using an Illumina HiSeq 2,500, NovaSeq 6,000, or MiSeq sequencer, with a reading length of 150 bp. In total, more than 6 GB of sequence data were generated for each transformant. The upstream and downstream genome sequences (500 bp) of the TISs were obtained using a comprehensive approach called TDNAscan³ searches against the genome database of WT C. magnum CAASZK4 (Sun et al., 2019). Using the GFF3 file of WT C. magnum as a reference genome organism, TDNAscan annotated all identified T-DNA insertions. Sequence homology searches were performed using NCBI protein database.⁴

Primers were designed to validate the TISs, T-DNA integration patterns, and deleted genes of Cm699, (Supplementary Table S2).

To determine the hygromycin B sensitivity of C. magnum, the CAASZK4 strain was incubated on PDA supplemented with various concentrations of hygromycin B. These results indicate that the growth of C. magnum was inhibited by 60 mg/L hygromycin (Supplementary Figure S1). To exclude the possibility of false-positive transformants, 70 mg/L was chosen for the subsequent selection of resistant transformants.

Using the modified ATMT protocol, we obtained a library containing 1,460 transformants of the watermelon pathogen C. magnum. On an average, 245 transformants were generated from 10⁸ conidia. Analysis of 40 randomly selected transformants harboring hygromycin B resistance revealed that all 40 transformants showed mitotically stable integrated T-DNA. These transformants, which were successively cultured for four generations on PDA medium without the selection marker hygromycin B, did not result in the loss of integrated T-DNA, as shown by the ability of all 40 transformants to grow on the screening medium containing hygromycin B and hph gene amplification analysis (Supplementary Figures S2, S3).

To effectively identify the T-DNA integration events in the mutant library, 40 randomly selected transformants and WT strains were subjected to genomic Southern blot hybridization. Genomic DNA digested using EcoRI and HindIII was detected with the digoxigenin (Dig)-labeled probe harboring 1,380 bp hph gene (Figure 1A). Of the tested transformants, 30 displayed single-site TISs, nine harbored two TISs, and one harbored more than two TISs. Representative selections of 30 transformants are shown in Figure 1B.

In total 1,460 transformants were analyzed for pathogenicity on detached leaves. Fourteen transformants were screened out for their strongly reduced virulence and further identified and analyzed on watermelon leaves. The three transformants of the 14 strains exhibited strongly reduced virulence in leaves assays. Interestingly, the three transformants (Cm699, Cm854, and Cm1078) caused no visible disease symptoms in watermelon leaves (Figure 2).

Three transformants (Cm699, Cm854, and Cm1078) exhibited significant morphology alterations compared to the WT strain (Figure 2). Of the three transformants, the hyphae of Cm699 showned severe deformity and pigment deposition compared to those of the WT strain (Figure 2); Cm699 and Cm1078 did not produce conidia, whereas Cm854 produced significantly fewer conidia than the WT strain (Figure 3A). In addition, the conidia germination rates of Cm854 were significantly lower than those of the WT strain (Figure 3B).

The growth rates of Cm699 (2.5) and Cm854 (10.8) were significantly lower than those of the WT strain (11.9), and there was no significant difference between the Cm1078 (11.6) and WT strains (Figure 4).

According to genomic Southern blot experiments, three transformants (Cm699, Cm854, and Cm1078) harbored single T-DNA integrations (Figure 1B). To identify the genomic loci of the transformants where T-DNA was inserted, TAIL-PCR was performed on three pathogenic transformants. PCR fragments displayed more specific and higher intensity bands in the third round than those in the first and second of TAIL-PCRs (Supplementary Figure S5). Thus, PCR amplicons resulting from the third round of TAIL-PCR were cloned and sequenced. Three genomic DNA sequences flanking the left and right border of the T-DNA ranging from 0.7 to 3.0 kb were obtained, respectively. In two transformants (Cm699 and Cm1078), comparison of the sequences flanking the left and right border of the T-DNA with the genome sequence of WT C. magnum revealed significant similarities (Table 2). However, the sequence flanking the T-DNA of Cm854 showed no significant similarity with the genomic sequence of WT C. magnum. A analysis of T-DNA insertions demonstrated that two transformants (Cm699 and Cm1078) had interrupted protein-coding genes (Table 3).

To further localize and characterize the TISs, whole-genome resequencing data from three transformants (Cm699, Cm854, and Cm1078) were utilized. Six T-DNA insertions in the three transformants were identified using TDNAscan (Table 4). However, all T-DNA insertions in the two transformants (Cm854 and Cm1078) were truncated (Table 5; Supplementary Figure S5). Furthermore, two T-DNA integration patterns, single-copy T-DNA with complete LB and RB and truncated tandem T-DNA repeats of LB or RB, were found in C. magnum (Figure 5). Using the local genome database of WT C. magnum as a reference genome, we found that T-DNA integration of transformant Cm699 occurred in the intron region starting 1,336-bp downstream of the start codon of the AP-2 complex subunit σ gene, and two genes were deleted, in the integration process. (Figure 5A). Based on whole-genome resequencing data, the two deleted genes were predicted to be the transcription initiation protein gene and hypothetical gene, respectively (Table 5). Furthermore, integration occurred within the promoter region 217-bp upstream of the start codon of the Myosin I gene in the transformant Cm854 (Figure 5B). In the Cm1078 transformant, the integration had in the exon region of two adjacent genes, one exon region starting 893-bp downstream of the start codon of WD-repeat protein retinoblastoma binding protein 4 gene (RBBP4), while the other exon region starting 4,812-bp upstream of the termination codon of 5′-3′exoribonuclease 1 gene (XRN1) (Figure 5C).