The WT strain of F. oxysporum f. sp. cubense used in the experiment was Foc4. The mycelium was maintained on potato dextrose Agar (PDA) medium and the conidial suspension was stored in 24% glycerol at −80 °C. The knockout and complement transformants were cultured on a medium containing 50 mg/mL hygromycin B or G418, respectively. E. coli DH5a, BL21 and A. tumefaciens GV3101, AGL1 were grown on a lysogeny broth (LB) medium containing corresponding antibiotics and were stored in 24% glycerol at −80°C. N. benthamiana was grown in a greenhouse with 8/16 (light/dark) per day at 24°C. Musa AAA, as warm-loving wet crops, was cultivated in a warm shed with a temperature of 30°C.

To investigate whether FoSsp1 can induce hypersensitivity response in N. benthamiana, Phusion Plus DNA polymerase (Thermo Fisher Scientific) was used to amplify the coding region of FoSSP1 and the coding region without signal peptide FoSSP1^ΔSP from the cDNA of Foc4 and cloned into the plant expression vector pBin-eGFP, respectively. The same sequence was amplified by PCR and cloned into a pSUC2 vector using EcoRI and XhoI restriction sites by ClonExpress^® II One Step Cloning Kit (Vazyme Biotech) for verifying the secretion activity of signal peptide (SP) (Oh et al., 2009). The encoding region and promoter sequence of FoSSP1 were amplified from WT genomic DNA and cloned into the pFL2 vector to obtain the complementary vector pFL2-HB (Chen et al., 2014).

To verify the expression of FoSSP1 during the process of the plant infection, roots of banana seedlings were inoculated with conidial suspension with a concentration of 10⁸/mL. Total RNAs were extracted from the inoculated roots (12, 24, 36, 48, 60, and 72 hpi). The relative expression of FoSSP1 in conidia was considered as the 0 hpi control. The extraction method refers to Eastep® Super Total RNA Extraction Kit (Promega), following the manufacturer's instructions. The cDNA was synthesized with the PrimeScript 1st strand cDNA Synthesis kit (Takara), according to the manufacturer's instructions.

A pair of specific (SSP1-RT-F/R) primers of the gene was used for qRT-PCR analysis with 100 ng cDNA as a template (Supplementary Table S1). The qRT-PCR was performed using the SuperReal PreMix Plus kit (Tiangen Biotech), following the manufacturer's instructions. The actin gene was used as an endogenous reference to normalize the expression of FoSSP1 in different periods and the relative expression levels of genes were calculated using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). Select one of the sets of data and arbitrarily set its relative expression to 1. The qRT-PCR conditions were as follows: an initial 95°C denaturation step for 3 min, followed by denaturation for 15 s at 95°C, annealing for 15 s at 57°C, and extension for 20 s at 72°C for 35 cycles (Wu et al., 2010). For the relative expression of PR genes in infected bananas, the actin gene (Degradi et al., 2021) of bananas was used as an endogenous reference to normalize the expression. One-way ANOVA was used to compare the significance of the difference between the two groups at the same time point.

The recombinant vectors pSUC2-FoSSP1 and pSUC2-FoSSP1^ΔSP were constructed and transferred into the YTK12 strain (Gu et al., 2011). The CMD-W medium (0.08% tryptophan dropout supplement, 0.65% yeast nitrogen base without amino acids, 2% sucrose, 0.1% glucose, and 2% agar) was used to screen the positive colonies from all the transformants. The successful transformant was cultured on a YPRAA medium to verify its secretion activity. To further test the secretion activity of the signal peptide, the reduction of 2, 3, 5-triphenyltetrazolium chloride (TTC) to insoluble red-colored 1, 3, 5-triphenylformazan (TPF) was monitored.

Gene replacement was carried out according to homologous recombination (Yang et al., 2018b). The upstream and downstream flanking sequences of FoSSP1 were cloned with primer pairs of UF/UR and DF/DR (containing the Hyg tag) [(Supplementary Table S1) Polynucleotide chain reaction primers used in this study], using the DNA of Foc4 as a template. The upstream and downstream of hygromycin resistant gene fragment was amplified with PCR primer pair of HYG-F/R and pEX2 vector as DNA template. Fusion fragment for transformation was generated by overlap PCR as described (Goswami, 2012). Purified fragments for protoplast transformation using Gel Extraction Kit (Omega) according to the manufacturer's instructions.

The hyphae were digested in enzymatic solution (25 g/L Driselase, 5 g/L Lysing Enzymes in 1.2 M KCl) and protoplasts were released until the concentration is 10⁷ protoplasts/ml (Yun et al., 2014). The DNA fragments or plasmids for transformation were mixed with protoplasts and incubated on ice for 20 min. Then, 1 ml PTC solution (0.6 M KCl, 50 mM Tris-HCl pH 8.0, 50 mM CaCl₂, and 40% PEG 4000) was added to the mixture and incubated at room temperature for 30 min. The mixture was transferred to a TB3 medium for further culture for 20 h. The transformed mixture was inoculated on a 15ml PDA medium containing 50 μg/ml hygromycin B or G418, then covered with PDA agar modified with 75 μg/ml hygromycin B or G418. Transformants harboring hygromycin B or G418 resistance were identified by PCR.

The recombinant vectors pBin-FoSSP1 and pBin-FoSSP1^ΔSP were transferred into Agrobacterium tumefaciens GV3101 by electroporation and cultured in LB containing 20 mg/ml rifampicin and 50 mg/ml kanamycin for 48 h. Bacteria were harvested and resuspended with injection buffer (1% MgCl₂·6H₂O, 1% MES, 150 μM acetosyringone) to the OD₆₀₀ of 1.0. The suspension was incubated at room temperature for 2–3 h to release the plasmids in the injection buffer. A plant of N. benthamiana at the stage of four leaves was selected and infiltrated the buffer from the back of the leaves (Wei et al., 2020). The empty pBin-eGFP vector was used as a negative control, and the vector containing BAX was used as a positive control. Leaves cell death symptoms were photographed both under natural light and under ultraviolet (wavelength 365 nm) at 6 days post-inoculation. Each assay was repeated three times.

Leaves were thoroughly ground in liquid nitrogen. The process of protein separation refers to the previous (Yang et al., 2012). The 20 μl total protein was separated with the 12.5% sodium dodecyl sulfate polyacrylamide gel electrophoresis (Shenggong Biotech) and transferred to the polyvinylidene difluoride (PVDF) membrane. Rabbit anti-GFP pAb (ABclonal) and rabbit anti-HA pAb (Shenggong Biotech) were used as primary antibodies and HRP Goat anti-rabbit IgG (ABclonal) as secondary antibodies.

To explore the growth rate, Fossp1, WT, and Fossp1-C were inoculated on the PDA plates, respectively, and cultured at 28°C for 6 days. The size of fungus mycelium was measured by crossing every day to calculate the mycelium growth rate. WT, Fossp1, and Fossp1-C were cultured in a PDB medium at 28°C. Up to the fourth day, the conidia numbers were counted daily using a hemocytometer.

WT, Fossp1, and Fossp1-C were cultured in a PDB medium for 4 days, and the spores were collected at least 10⁸ conidia/ml. The roots of banana seedlings were cleaned and root wounds were made with an art knife, then immersed in the prepared spore solution for 30 min. The seedlings were transplanted into flowerpots. And 50 mL conidial suspension was poured into each pot.

To evaluate the disease index, the corm of banana seedlings at 20 dpi was cut to measure the lesion area. The lesion area was used for disease grade, disease grade for each plant ranging from 0 to 4 (0, no lesion in corm; 1, 1–10% lesion area in corm; 2, 11–30% lesion area in corm; 3, 31–50% lesion area in corm; 4, more than 50% lesion area in corm). The disease index of infected banana seedlings was calculated as follows formula: Disease index (%) = [∑ (grade × number of plants in that grade)/ (4 × total number of assessed plants)] × 100% (Zhang et al., 2019).

Agrobacterium containing recombinant vectors and pBin-FoSSP1^ΔSP was re-suspended in buffer solution at OD₆₀₀ of 0.001. The syringe was used to infiltrate Agrobacterium solutions into the tobacco leaves. The leaves were photographed and observed under a laser confocal microscope 2 days post infiltration. DAPI staining was conducted as previously described (Tarnowski et al., 1991). The empty vector was set as the negative control.

The accumulation of H₂O₂ was detected by staining tobacco leaves with 3, 3'-diamino benzidine (DAB). The leaves were immersed in A. tumefaciens for 3 days and stained in DAB for 12 h in the dark. Then the leaves were destained in 95% ethanol until translucent and photographed (Dong and Chen, 2013).

For callose staining, tobacco leaves were boiled and decolorized in 95% ethanol, and then immersed in 1% aniline blue solution in the dark for at least 2 h. The deposition of callose was detected under the microscope.

A small secreted protein of plant pathogens is characterized by an N-terminal signal peptide, no transmembrane domain, rich-in cysteine, and no more than 300-amino acids (Zhang et al., 2020). Here, SignalP (Nielsen, 2017) (https://services.healthtech.dtu.dk/service.php?SignalP-5.0), TMHMM (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0) and Big-PI Predictor (https://mendel.imp.ac.at/gpi/fungi_server.html) were used to predict the small secreted proteins of F. oxysporum f. sp. Cubense Race 4 (Foc4). More than 1% of cysteine is considered rich in cysteine. In total, 106 putative secreted proteins were obtained that met the above characteristics. Among them, 11 proteins were selected randomly for transient expression in tobacco (Supplementary Table S2). Small secreted proteins screened in this study). The encoding genes were constructed into a recombinant pGR107 vector and transiently expressed in N. benthamiana leaves by A. tumefaciens-mediated transformation. The results showed that FOIG_06406 (Gene ID: 42031581) which encodes a 145-amino acids protein can induce necrotic cell death in N. benthamiana (Figure 3A). We named it FoSSP1 (Small Secreted Protein) in this study. A BLASTp search against the NCBI Nr database was conducted using the FoSsp1 protein sequence (NCBI Reference Sequence: XP_031064189.1) as a query, results indicated that FoSsp1 homologs are conserved in the Fusarium genus and many other higher Ascomycetes. Interestingly, FoSsp1 homologs were also found in Pterula gracilis, which belongs to Basidiomycetes. These results indicate that FoSsp1 was conserved in fungi (Figure 1).

Fungal secreted proteins usually participate in pathogen infection. To explore the relative expression of FoSSP1, qRT-PCR was conducted in banana roots infected with Foc4. Compared with the expression of FoSSP1 in conidia (0 hpi, hour post-inoculation), the expression of FoSSP1 was significantly up-regulated at 60 and 72 hpi (hours post-inoculation). There was no significant difference between the expression levels in other periods and the gene expression levels in spores (Figure 2). This result indicates that FoSSP1 is induced in plant infection, implying its function in the plant-pathogen interaction.

To dissect the effect of FoSsp1 on plant defense, transient expression of FoSsp1 in N. benthamiana was conducted. The cell death triggering protein Bax (Gene ID: 581) (Lacomme and Santa Cruz, 1999), was set as a positive control. Given that signal peptide is often thought to be cleaved to form mature proteins (Owji et al., 2018), recombinant vectors of pBin-eGFP-FoSSP1 and pBin-eGFP-FoSSP1^ΔSP were constructed respectively. Results indicate that Bax, pBin-eGFP-FoSSP1, and pBin-eGFP-FoSSP1^ΔSP induced cell death in leaves of N. benthamiana after 5 days post agroinfiltration, whereas the EV (empty vector) did not (Figure 3A). Proper protein expression in N. benthamiana leaves was further confirmed by Western Blot analysis (Figure 3B). The experiment was repeated three times with the same results. The above results showed that FoSsp1 could induce cell death in N. benthamiana.

Split-marker approach was used to generate the deletion mutant of FoSSP1 in Foc4. The nucleotide sequence of a segment containing the FoSSP1 gene (NW_022158695.1: 156,986-157,890 nt) was replaced by a hygromycin-resistance gene in Foc4. The resulting transformants harboring the hygromycin B resistance were further confirmed by PCR. In total, five strains of Fossp1 deletion mutant were generated and FT1 was chosen for further experiments [(Supplementary Figure S1) PCR Test of Fossp1 deletion mutants]. To generate the complement strain of the Fossp1 mutant, the promoter and the encoding region of FoSSP1 (NW_022158695.1: 157,201-157,734 nt) were constructed into the pFl2 plasmid and transformed into the Fossp1 deletion mutant by PEG-mediated protoplast transformation. The resulting transformants harboring neomycin (G418) resistance were screened by PCR. In total, six complement strains were generated.

To examine the growth rate, wild-type strain of Foc4, Fossp1 deletion mutant, and complement strain of Fossp1-C were cultured on PDA agar plates for 6 days. The results showed that no significant difference in growth rate among WT (0.59 ± 0.01 cm/d), Fossp1-C (0.61 ± 0.01 cm/d) and Fossp1 deletion mutant (0.59 ± 0.01 cm/d) (Figures 4A,B).

The conidia production of Foc4, Fossp1-C, and Fossp1 deletion mutant were calculated in PDB medium 2 days after inoculation. Interestingly, the conidial yield of Fossp1 mutant (5.61 ± 0.35 × 10⁵ conidia/ml) was significantly higher than that of WT strain (5.04 ± 0.24 × 10⁵ conidia/ml) and complement strain Fossp1-C (5.05 ± 0.21 × 10⁵ conidia/ml) (Figure 4C). These results suggest that the FoSSP1 gene regulates the sporulation of Foc4.

To analyze the pathogenicity of FoSSP1, it was knocked out and complementary successfully (Supplementary Figure S1). The knockout mutant Fossp1, complement strain Fossp1-C, and the WT Foc4 on banana seedlings (Cavendish banana var. Brazil (Musa AAA) was tested. The conidia suspension (1 × 10⁸ conidia/ml) of Fossp1 mutant, complement strain Fossp1-C, and WT Foc4 were inoculated to the banana plantlets. At 10 dpi (days post inoculation), no symptoms presented on leaves and external bulbs of the Fossp1, Fossp1-C, and Foc4 (Figure 5A upper and middle panel). However typical discoloration internal of bulbs appeared when observed in the vertical section of the bulbs of Fossp1 deletion mutant, but not of WT and complement strain (Figure 5A, lower panel). At 20 dpi, typical yellowing leaves of seedlings infected by Foc4 were observed. In comparing, two or more yellowing leaves of each seedling inoculated by the Fossp1 deletion mutant, while of the wild-type and the complement strains yellowing leaves up to one of each seedling (Figure 5B). To quantify the disease severity of the infected banana seedlings with different Foc4 strains, the disease index of infected banana seedlings at 20 dpi was calculated. The disease index of banana plants infected with Fossp1 deletion mutant was significantly increased than that of the WT strain Foc4 and complement Fossp1-C at 20 dpi (Figure 5D). When observed banana seedlings at 30 dpi, all plants infected by Foc4 displayed leaf yellowing, even the death of Fossp1 (Figure 5C, upper panel). In addition, the pseudostem and root displayed typical discoloration of infected seedlings (Figure 5C, lower panel). Discoloration length of pseudostem was measured at 30 dpi, in comparison to WT Foc4 (4.15 ± 0.17 cm) and the complement strain (7.55 ± 0.31 cm), discoloration length of pseudostem of Fossp1 (4.20 ± 0.30 cm) increased by about 50%. These results indicated that deletion of FoSSP1 enhances the virulence of Foc4 in banana seedlings.

Deletion of FoSSP1 in Foc4 leads to the enhancement of sporulation capability. We hypothesized that FoSSP1 might regulate Foc4 sporulation-related genes. Firstly, RNA of mycelia grown in a PDB medium for 2 days was extracted. Then, the transcription levels of the sporulation-related genes in WT and Fossp1 deletion mutant were tested by quantitative reverse transcription PCR (qRT-PCR). Compared with the WT, the key genes in the sporulation pathway, BrlA, Fgb, wet, and AbaA were significantly down-regulated (Figure 6A).

Plant pathogenesis-related genes play a crucial role in plant defense against the pathogen (Joshi et al., 2021). Deletion of Fossp1 to analyze differential expression of pathogenesis-related genes, the roots of banana seedlings were inoculated with a spore suspension of Foc4 and Fossp1, respectively, and total RNA was extracted 3 days post-inoculation to detect transcriptional levels of pathogenesis-related genes. RT-qPCR showed that the transcription level of PRs gene, PR1b, PR2b, PR5, and PR10 in root infected by Fossp1 was significantly lower than that did the WT (Figure 6B). The results of two qRT-PCR showed that FoSSP1 induced the expression of host defense genes during the pathogen infection of roots. Furthermore, in the process of conidiation, FoSSP1 participates in the central regulatory pathway controlling conidiation.

Secreted proteins can be divided into typical secreted proteins and atypical secreted proteins according to the presence or absence of signal peptides. A total of 23 amino acids (aa 1-23) at the N-terminal of FoSsp1 were predicted to be a signal peptide by SignalP5.0. The secretory function of the FoSsp1 signal peptide was verified using the yeast secretion system (Gu et al., 2011). The YTK12 strain was transformed with the reconstructed pSUC2 vector to express the signal peptide of pSUC2-FoSsp1 and pSUC2- FoSsp1^ΔSP. The signal peptide of a known secreted protein, Avr1b, was used as a positive control (Tian et al., 2021). Untransformed YTK12 and empty pSUC2 vectors were used as negative controls. Similar to the Avr1b signal peptide, the FoSsp1 but not the FoSsp1^ΔSP or the empty vector, allowed YTK12 to grow on the YPRAA medium (Figure 7A). The strain transformed with the FoSsp1 with its signal peptide displayed secreted invertase enzyme activity that catalyzed the reduction of TTC to a red-colored compound (Figure 7A). These results suggest that FoSsp1 is a secreted protein, and its signal peptide is required for its secretion.

To explore the subcellular localization of FoSsp1 in plant cells, the recombinant vector pBin-eGFP-FoSSP1 and pBin-eGFP-FoSSP1^ΔSP were transiently expressed in tobacco leaves, and pBin-eGFP was used as a negative control. The plant nucleus was labeled with DAPI (2-(4-Amidinophenyl)-6-indolecarbamidine dihydrochloride) staining. When observed under the microscope, bright green fluorescence in the leaf cell nucleus of tobacco was overlapped with the blue fluorescence of the DAPI staining (Figure 7B). Interestingly, a weak fluorescence signal of pBin-eGFP-FoSSP1 was also distributed in extracellular and intracellular tobacco cells (Supplementary Figure S2). The result showed that FoSsp1 localized in the nucleus of a tobacco leaf cell.

Due to the increased virulence phenotype of the Fossp1 deletion mutant, we speculate that FoSsp1 is an elicitor that can stimulate the plant immune response. Therefore, the accumulation of H₂O₂ and the callose deposition in tobacco were detected after A. tumefaciens infiltration. FoSsp1 accumulates more H₂O₂ and callose than GFP control (Figure 8A). The results showed that FoSsp1 can activate the defense response of host plants, promoting oxygen burst and callose deposition.

To further explore the defense response of N. benthamiana activated by FoSsp1, the expression of defense-related genes was detected using qRT-PCR. The genes pathogenesis-related protein 1B (PR1b), pathogenesis-related protein 2B (PR2b), ethylene-responsive transcription factor 1 (ERF1), and linoleate 9S-lipoxygenase 5 (LOX) related to salicylic acid, ethylene, and jasmonic acid signal pathways were detected (Wang et al., 2000; Lorenzo et al., 2003; Wu et al., 2018). There were statistically significant differences in the expression of these genes. From the results, it is clear that PR1b and LOX showed higher expression levels at 48 hpi, while ERF1 expression significantly increased at 24 hpi compared with control (Figure 8B).