Loss of glucosylceramide synthase impairs the growth and virulence of Fusarium oxysporum f. sp. cubense

Glucosylceramides are a class of membrane lipids that serve as vital structural and signaling molecules in eukaryotes. In this study, we explored the function of FocGCS, a glucosylceramide synthase (GCS) in Fusarium oxysporum f. sp. cubense tropical race 4 (Foc TR4) that causes Fusarium wilt in banana plants. FocGCS is highly expressed in germinating conidia and during early infection stage of Foc TR4. Disruption of FocGCS resulted in severely retarded vegetative growth, reduced conidiation, and production of morphologically abnormal conidia. Sphingolipid profiling revealed that the FocGCS null mutant lacks glucosylceramide. Pathogenicity assays on banana plants revealed substantial loss of virulence in the FocGCS null mutant. Moreover, biochemical analyses indicated that FocGCS is involved in cell wall integrity but is not required for oxidative and osmotic stress tolerance in Foc TR4. Transcriptome analysis suggested that disruption of FocGCS strongly affects transmembrane transport in Foc TR4. Our findings show that GCS is essential for normal fungal growth and pathogenesis in Foc TR4.


Background
The soil-borne anamorphic fungus Fusarium oxysporum f. sp. cubense (Foc) colonizes and blocks the vascular system of host plants, causing Fusarium wilt (also known as Panama disease), the most destructive fungal disease on banana plants worldwide (Ploetz 2006;Fourie et al. 2009). In general, Foc-infected banana plants die before they produce bunches of fruit, and this has led to reduced banana yields in Asia, Africa, Australia, and the tropical Americas in the past few decades (Li et al. 2011;Liu et al. 2019a). Based on its host types, Foc can be classified into four races (Ploetz 2006;Dita et al. 2010). Foc tropical race 4 (Foc TR4) is highly pathogenic to almost all banana cultivars and has destroyed thousands of hectares of Cavendish bananas in tropical and subtropical countries (Molina et al. 2009;Zuo et al. 2015). To date, there are no effective biological, chemical, and cultural measures to manage this disease (Ploetz 2015).
Glucosylceramides (GlcCers) are common membrane sphingolipids in most eukaryotic organisms. Glucosylceramide synthase (GCS) catalyzes the transfer of a sugar residue from UDP-glucose to ceramide to form GlcCer (Wang et al. 2021). Although the GCS gene is absent in Saccharomyces cerevisiae, it is present in most of the other fungal species studied so far, such as phytopathogens Magnaporthe grisea, Colletotrichum gloeosporioides, and Fusarium graminearum (Leipelt et al. 2001;Dickson and Lester 2002;Ramamoorthy et al.

Open Access
Phytopathology Research  Poeta et al. 2014;Huang et al. 2019). Several studies have suggested that GlcCer modulates membrane physical properties and physiological functions in some fungal pathogens of humans to facilitate fungal growth, spore germination, virulence, and hyphal development and differentiation (Levery et al. 2002;Warnecke and Heinz 2003;Thevissen et al. 2004;Ramamoorthy et al. 2007;Zhu et al. 2014). In the human fungal pathogen Cryptococcus neoformans, GlcCer is essential for fungal growth and is a key virulence factor (Rittershaus et al. 2006). The GlcCer of C. neoformans is a fungal antigen that plays a role in cell wall synthesis and yeast budding. The antibodies against C. neoformans GlcCer produced by patients with cryptococcosis inhibit hyphal growth and cell division in vitro (Rodrigues et al. 2000). In Candida albicans, GlcCer biosynthesis is an important determinant of virulence in a murine model (Noble et al. 2010). In Aspergillus nidulans and Aspergillus fumigatus, inhibiting GlcCer synthesis using 1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP), an inhibitor of GCS in mammalian cells, adversely affected fungal growth and spore germination (Levery et al. 2002). In the phytopathogen Penicillium digitatum, deletion of PdGcs1 resulted in impaired virulence on citrus fruits (Zhu et al. 2014). In C. gloeosporioides, CgGCS-knockout mutants almost completely lost the ability to invade natural host plants tomato (Solanum lycopersicum) and mango (Mangifera indica) (Huang et al. 2019). However, the role of GCS in pathogenesis of F. graminearum is host-dependent (Ramamoorthy et al. 2007). The results of these studies indicate that GlcCers are involved in the growth and virulence of various fungal pathogens.
Additionally, antifungal defensins, which are cationic antimicrobial proteins and part of plant innate immunity, bind to distinct GlcCers in the membranes of fungi and inhibit fungal growth. For instance, the defensin RsAFP2 produced by Raphanus sativus interacts specifically with GlcCers from Pichia pastoris and C. albicans, resulting in growth arrest of the pathogens (Thevissen et al. 2004). The defensin MsDef1 from Medicago spp. interacts with GlcCer produced by F. graminearum, inhibiting fungal growth (Ramamoorthy et al. 2007). Together, these studies suggest that fungal GlcCer and GCS are promising biologically active targets for new antimicrobials (Nimrichter and Rodrigues 2011).
In this study, we report the identification and functional analysis of GCS in Foc TR4. We found that FocGCS is a key regulator of fungal growth and conidiation and is indispensable for cell wall integrity and full virulence of Foc TR4.

Characterization of FocGCS
Using BLASTp, we found a putative GCS-encoding gene (XP_031069237.1) in Foc TR4, which is homologous to the GCS genes in Homo sapiens (NP_003349.1), C. albicans (KGU07664.1), and F. graminearum (XP_011324566.1), with 23.73%, 34.41%, and 68.05% amino acid identity, respectively (Additional file 1: Table S1). Analysis of the amplified 1815-bp genomic and 1578-bp cDNA sequences indicated that FocGCS contains a single 51-bp intron, and the entire coding sequence of the gene is 1581 bp, encoding a protein of 526 amino acids with an estimated molecular mass of 58.8 kDa.
The predicted FocGCS protein contains a conserved Glyco_tranf_21 domain, an N-terminal transmembrane domain, and three C-terminal transmembrane domains (Additional file 2: Figure S1a). These domains, including the D1, D2, D3, and Q/RXXRW motifs (Additional file 2: Figure S1b), are essential for the enzymatic activity of mammalian GCS; histidine 193 is conserved in the Glyco_tranf_21 domain (Leipelt et al. 2001). Phylogenetic analysis indicated that FocGCS is closely related to GCSs from F. graminearum, Pyricularia oryzae, Beauveria bassiana, C. gloeosporioides, and Verticillium dahliae, with 68.05%, 47.32%, 50.75%, 53.83%, and 50.64% amino acid sequence identity, respectively (Fig. 1a), indicating that FocGCS is well conserved in fungi. Based on amino acid sequence similarity of GCSs, the most distantly related fungus of Foc TR4 in the phylogenetic tree is Rhizoctonia solani (30.95% identity). FocGCS has 23.73% and 11.14% amino acid sequence identity to GCSs from Homo sapiens and Arabidopsis thaliana, respectively.

FocGCS is highly expressed at different developmental stages and during early infection stage of Foc TR4
To examine the expression of FocGCS at different developmental stages of Foc TR4, we constructed the FocGCS-GFP fusion cassette, which expresses FocGCS fused to green fluorescent protein (GFP) under the control of the FocGCS promoter. When expressed in the ΔFocGCS-C-GFP strain described below, the FocGCS-GFP protein was observed in conidia and hyphae, especially in the Golgi, endoplasmic reticulum, and vacuole, whereas fluorescent signals produced by the Foc-GFP strain, which expresses an un-fused GFP, was observed throughout conidia and hyphae (Additional file 2: Figure S2).
We investigated the transcript level of FocGCS in mycelia, conidia, germinating conidia, and Foc TR4-infected banana roots at different time points of early infection by quantitative real-time PCR (qPCR). Compared with that in mycelia, FocGCS transcript level showed a several-fold increase during conidial germination (Fig. 1b).
In addition, in Foc TR4-inoculated plants, the transcript level of FocGCS increased gradually during the early infection stages until 48 h post-infection (hpi) and then decreased (Fig. 1b, c).

Generation of FocGCS-knockout mutant and complemented strain
To explore the role of FocGCS, we generated FocGCSknockout mutant using a targeted gene replacement strategy. Hygromycin B 4-O-kinase-coding gene (HPH) driven by the TrpC promoter was transformed into the Foc TR4 strain II5 via homologous recombination to replace the GCS gene (Additional file 2: Figure S3a). The putative mutants were screened by PCR with the primer pairs listed in Additional file 1: Table S2. PCR amplification of the 5′ and 3′ terminus of FocGCS with the primer pairs 1/2 and 3/4 produced a 1983-bp and a 1951-bp fragment, respectively, in the ΔFocGCS mutant, but not in the wild-type (WT) strain, suggesting correct integration of HPH (Additional file 2: Figure S3b). qPCR analysis detected no GCS transcript in the ΔFocGCS mutant (Additional file 2: Figure S3c). Southern blotting of genomic DNA using a 502-bp FocGCS-specific probe revealed the presence of a 7.5-kb XbaI fragment in the WT strain, but not in the ΔFocGCS mutants ( Fig. 2a), demonstrating targeted disruption of the FocGCS gene in the mutants.
Complementation of the ΔFocGCS mutant was performed with the FocGCS-GFP construct described above. This construct was introduced into a representative ΔFocGCS strain to produce the complemented ΔFocGCS-C-GFP strain. Several neomycin-resistant transformants were selected and verified by PCR amplification, which produced a 980-bp fragment identical to that obtained from the WT strain but absent from the ΔFocGCS mutant (Additional file 2: Figure S3b). qPCR analysis revealed that the GCS transcript level in the complemented strain was similar to that in the WT strain (Additional file 2: Figure S3c).

FocGCS is required for GlcCer biosynthesis in Foc TR4
To examine whether targeted disruption of FocGCS affects the biosynthesis of sphingolipids, we compared the contents of GlcCers, hydroxyceramides (hCers), ceramides (Cers), and long chain bases (LCBs) in hyphae of the WT and mutant strains using high-performance liquid chromatography-mass spectrometry/mass spectrometry (HPLC-MS/MS). The GlcCers d19:2 g18:0, d19:2 g18:1, d19:2 g20:0, and d19:2 g24:0 were almost undetectable in the ΔFocGCS mutant (Fig. 4a, d). By contrast, the amounts of hCers d19:2 h18:0, d19:2 h18:1, d19:2 h20:0, and d19:2 h24:0 were significantly higher in the mutant strain than those in the WT strain ( Fig. 4b and Additional file 2: Figure S4). As shown in Fig. 4c, the total amount of hCers was approximately sixfold greater in the mutant strain than that in the WT strain, while the total amount of GlcCers was much lower in the mutant strain than that in the WT strain. There was no significant difference in accumulation levels of LCBs and Cers between the WT and mutant strains. These results indicate that targeted disruption of FocGCS resulted in increased accumulation of hCers whereas reduced accumulation of GlcCers, suggesting that FocGCS functions as a glucosylceramide synthase in Foc TR4.

FocGCS contributes to the full virulence of Foc TR4
To determine whether disruption of FocGCS affects the virulence of Foc TR4, pathogenicity test was conducted on Cavendish banana plantlets at the 6-7 leaf stage through root infection assays. The WT-and ΔFocGCS-C-GFP-inoculated plants showed severe disease symptoms at 4 weeks post-inoculation. In comparison, the degree of browning in inner rhizome of ΔFocGCS-inoculated plants was significantly reduced (Fig. 5a). Disease severity in ΔFocGCS-inoculated plants was dramatically lower than that in plants inoculated with other tested strains (Fig. 5b).
To determine whether FocGCS knockout affects fungal growth in planta, we quantified fungal biomass of the WT, ΔFocGCS, and ΔFocGCS-C-GFP strains in root tissues. The fungal biomass in banana roots inoculated with Southern blot identification and growth phenotype of the ∆FocGCS mutant. a Southern blot analysis of genomic DNA extracted from the WT and three putative ∆FocGCS strains and digested with XbaI. b Colony morphology of the WT, ∆FocGCS, and ∆FocGCS-C-GFP strains on PDA plates. Photos were taken after incubation at 28 °C for 5 days. c Colony diameters of the WT, ∆FocGCS, and ∆FocGCS-C-GFP strains grown on PDA plates. The mean values ± SD were calculated from three independent experiments, and asterisks indicate significant differences from the WT group (Student's t-test, ***P < 0.001) ΔFocGCS was significantly reduced compared with that in WT-inoculated roots at 5, 10, and 15 days post-inoculation (dpi) (Fig. 5c). These data suggest that knockout of FocGCS retarded mycelial extension rate of Foc TR4 within banana roots and reduced the virulence of the pathogen on banana plants.

FocGCS affects cell wall integrity of Foc TR4
GlcCers are essential structural components of fungal cell wall, plasma membrane, and intracellular vesicles (Ramamoorthy et al. 2007;Fernandes et al. 2016). Therefore, we analyzed the sensitivity of the WT, ΔFocGCS, and ΔFocGCS-C-GFP strains to cell-wall-damaging agents (0.2% Calcofluor white (CFW) or 0.2% Congo red (CR)). After 5 days of incubation on PDA plates supplemented with or without the agent, the mutant strain showed increased CFW and CR resistance compared with the WT and complemented strains (Fig. 6a). The growth inhibition rate of the ΔFocGCS mutant decreased by 25% and 19% on CFW and CR plates, respectively, compared with that of the WT strain (Fig. 6b), indicating that the mutant was more resistant to cell-wall-perturbing agents.
Considering that chitin is also the major component of fungal cell wall, and CFW and CR inhibit fungal cell wall assembly by binding to chitin, the chitin contents in the WT, ΔFocGCS, and ΔFocGCS-C-GFP strains were also compared (Schoffelmeer et al. 1999). The results showed that the chitin content in the mutant was 33% and 27% higher than that in the WT and complemented strains, respectively (Fig. 6c). Apparent changes in cell wall structure were observed in ∆FocGCS strains under transmission electron microscopy (TEM), including thickening of the cell wall and formation of more fungal-type vacuole (Fig. 6d), and the thickness of the cell wall in ΔFocGCS strain increased by 22% compared with that of the WT strain (Fig. 6e). These results indicate that FocGCS is involved in cell wall integrity and chitin biosynthesis of Foc TR4. The mean values ± SD were calculated from three independent experiments. c Conidia production of the indicated strains incubated for 5 days on PDA plates. d Conidial germination rate of the strains incubated on YEPD broth for 12 h. Germination of 100 conidia was randomly examined. Asterisks indicate significant differences from the WT group (Student's t-test, ***P < 0.001)

FocGCS is not involved in the tolerance of Foc TR4 to oxidative and osmotic stresses
The production of reactive oxygen species is necessary for full virulence of Foc TR4 and plant defense (Qi et al. 2013). We therefore investigated whether FocGCS is involved in oxidative stress response by exposing the WT, ΔFocGCS, and ΔFocGCS-C-GFP strains to 0.5 mM H 2 O 2 (Fig. 7a). The difference in mycelial growth rate between these three strains was not significant (nearly 65% inhibition for all the strains), indicating that FocGCS is not essential for tolerance to oxidative stress in Foc TR4 (Fig. 7a, b).
Because GlcCer mainly functions as structural component of cell membrane, we speculated that GlcCer defect might influence the response to osmotic stress. We therefore examined the growth of the WT, ΔFocGCS, and ΔFocGCS-C-GFP strains on PDA plates supplemented with NaCl or sorbitol. No differences in colony growth were observed among the three strains incubated on the plates containing 1 M NaCl or 1 M sorbitol (Fig. 7a, c). These findings indicate that FocGCS is not necessary for osmoregulation in Foc TR4.
GlcCers have been reported to be involved in regulating pH tolerance (Rittershaus et al. 2006;Ramamoorthy et al. 2007). However, the WT, ΔFocGCS, and ΔFocGCS-C-GFP strains exhibited similar pH-dependent growth trends, with the ΔFocGCS mutant showing greater inhibition at pH 10 (Fig. 7d, e). These results suggest that GlcCer in Foc TR4 does not broadly participate in environmental pH regulation but may influence alkali tolerance.

Comparative transcriptome analysis between the Foc TR4 WT strain and the ΔFocGCS mutant
To comprehensively investigate the difference in gene expression between the ΔFocGCS mutant and the WT The mean values ± SD were calculated from three independent experiments, and asterisks indicate significant differences from the WT group (Student's t-test, **P < 0.01, ***P < 0.001) strain, we performed comparative transcriptome analysis. The results showed that 571 genes exhibited at least a twofold difference in expression between the WT and mutant strains (Additional file 3: Table S3). Functional classification of differentially expressed genes (DEGs) was further conducted to explore the transcriptome regulation pattern. DEGs matching well-characterized proteins or proteins with putative functions were grouped according to their Gene Ontology (GO) categories, as summarized in Additional file 3: Table S3. Most DEGs identified in this study were assigned to different biological and molecular function categories. These DEGs encode proteins that are predicted to be involved in the oxidation and reduction, ion binding, and transmembrane processes (Fig. 8a, b). 'Transmembrane transport' and 'Transmembrane transporter activity' were the most significantly enriched GO terms for these DEGs (Fig. 8c), implying that interrupting transmembrane transport in the mutant strain may account, in part, for the differences between the ΔFocGCS mutant and the Foc TR4 WT strain.

Discussion
GlcCers are important membrane lipids in animals, plants, and fungi. In this study, we identified a GCS gene, termed FocGCS, in Foc TR4, with the goal to determine its role in fungal growth, development, and virulence. FocGCS shares substantial sequence identity with homologs in other fungi (Additional file 2: Figure S1b and Additional file 1: Table S1) (Leipelt et al. 2001). Our findings demonstrate that FocGCS encodes a bona fide GCS whose product, GlcCer, is essential for regulating fungal growth, conidiation, conidial germination, cell wall integrity, and virulence of Foc TR4.
Unlike in S. cerevisiae, GlcCers are produced in filamentous and dimorphic fungi, such as C. albicans, F. graminearum, and P. digitatum (Dickson and Lester 2002;Ramamoorthy et al. 2007;Noble et al. 2010;Zhu et al. 2014). Disruption of GCS can reduce mycelial growth and virulence of pathogenic fungi (Rittershaus et al. 2006;Ramamoorthy et al. 2007;Zhu et al. 2014;Huang et al. 2019). In our study, deletion of FocGCS severely retarded the growth of Foc TR4 (38% reduction in growth rate compared with the WT and complemented strains) (Fig. 2b, c), indicating that GCS is essential for normal growth of the pathogen. We speculate that growth deficiency in Foc TR4 may contribute to its impaired pathogenesis. Furthermore, knockout of FocGCS reduced the virulence of Foc TR4 on host plants, consistent with the previous studies (Zhu et al. 2014;Huang et al. 2019). The characterization of FocGCS knockout mutant in this study provides insight into the functional roles of this enzyme in vivo. Deletion of the putative glycosyltransferase gene in F. graminearum, P. digitatum, and C. gloeosporioides resulted in a complete loss of GlcCer (Ramamoorthy et al. 2007;Zhu et al. 2014;Huang et al. 2019). Our results indicate that knockout of FocGCS also had a dramatic effect on the synthesis of GlcCer, resulting in nearly complete GlcCer deficiency (Fig. 4a). Meanwhile, the accumulation of two direct substrates of GCS, d19:2 h18:1 and d19:2 h18:0, was substantially increased in the mutant strain (Fig. 4b). This confirmed that FocGCS indeed encodes a GCS that catalyzes the synthesis of GlcCer. In the ΔFocGCS mutant, inhibition of GlcCer was accompanied by severe growth defects, along with reduced production of macroconidia and formation of conidia with apparently smaller size compared with the WT strain. These results are consistent with those reported for A. fumigatus, A. nidulans, F. graminearum, and C. albicans (Levery et al. 2002;Ramamoorthy et al. 2007;Noble et al. 2010).
Previous reports also suggested that GlcCer affects fungal pathogenesis (Ramamoorthy et al. 2007;Huang et al. 2019). In F. graminearum, the growth of FgGCS1knockout mutant that lacks GlcCer was retarded, but its ability in causing disease was host-dependent (Ramamoorthy et al. 2007). Pathogenicity assays indicated that the CgGCS deletion mutant of C. gloeosporioides almost completely lost the ability to invade tomato and mango hosts (Huang et al. 2019). In this study, the disease symptom in ΔFocGCS-inoculated plants was significantly reduced compared with that in WT-inoculated plants (Fig. 5), indicating that GlcCer deficiency attenuated the virulence of the pathogen. In addition, our studies showed that disruption of FocGCS affected cell wall integrity, which has been reported to be involved in Fig. 6 Cell wall integrity analysis of the WT, ∆FocGCS, and complemented strains. a Colony morphology of the indicated strains on PDA plates supplemented with 0.2% Calcofluor white (CFW) or 0.2% Congo red (CR). Photos were taken after incubation at 28 °C for 5 days. b The growth inhibition rate based on a. The inhibition rate was calculated by comparing the colony diameter under stress conditions with that under normal conditions. c Cellular chitin levels of the indicated strains measured by colorimetric determination of β-(1-4)-linked N-acetylglucosamine (GlcNAc). DW, dry weight. d Transmission electron micrographs of the WT and ∆FocGCS strains. CW, cell wall; M, mitochondrion; N, cell nucleus; V, vacuole. Scale bars = 0.2 μm. e Cell wall thickness of the indicated strains measured by 100 spores. The mean values ± SD were calculated from three independent experiments, and asterisks indicate significant differences from the WT group (Student's t-test, *P < 0.05, **P < 0.01, and ***P < 0.001) virulence (Kong et al. 2012). Our transcriptome analysis showed that there was no difference in the expression of major virulence genes such as Secreted In Xylem (SIX), FocCP1, and FocPGC4 genes (Additional file 3: Table S3) (Czislowski et al. 2018;Liu et al. 2019a;Dong et al. 2020). Functional classification of DEGs showed that FocGCS also affected transmembrane transport of Foc TR4, but whether transmembrane transport is related to virulence needs to be determined.
Recently, GlcCers have been reported to be involved in the regulation of pH tolerance in phytopathogens. GlcCer is critical for C. neoformans to grow in a neutral/alkaline but not in an acidic extracellular environment (Rittershaus et al. 2006). The GCS null mutant of F. graminearum exhibits highly reduced radial growth on solid medium, and its ability to grow at high pH is severely impaired (Ramamoorthy et al. 2007). The results of this work suggested that GlcCer in Foc TR4 does not Fig. 7 Stress tolerance assays of the WT, ∆FocGCS, and complemented strains. a Colony morphology of the indicated strains on PDA plates supplemented with 0.5 mM H 2 O 2 , 1 M NaCl, or 1 M sorbitol. Photos were taken after incubation at 28 °C for 5 days. b Colony growth inhibition incubated with 0.5 mM H 2 O 2 . c Colony growth inhibition incubated with 1 M NaCl or 1 M sorbitol. The inhibition rate was calculated by comparing the colony diameter under stress conditions with that under normal conditions. d Colony morphology of the indicated strains on PDA plates with different pH values. Photos were taken after incubation at 28 °C for 5 days. e Colony growth inhibition under different pH conditions. The inhibition rate was calculated by comparing the colony diameter under different pH conditions with that under condition of pH = 6.0. The mean values ± SD were calculated from three independent experiments, and asterisks indicate significant differences from the WT group (Student's t-test, **P < 0.01) broadly participate in environmental pH regulation but may have a small effect on alkali tolerance. (Fig. 7d, e). However, in the GlcCer-deficient Foc TR4 strain, whether the environmental pH regulation influences its pathogenesis remains to be determined.
Several studies have provided evidence that Glc-Cer may be important for cell wall organization in pathogenic fungi. In C. neoformans, GlcCer is mainly localized in the cell wall and accumulates mostly at the budding site of dividing cells (Rittershaus et al. 2006). The loss of GlcCer in F. graminearum results in greatly enhanced resistance to cell-wall-perturbing agents (Ramamoorthy et al. 2007;Rittenour et al. 2011). Similarly, we found that targeted disruption of FocGCS affects the cell wall integrity of Foc TR4, as the ΔFocGCS mutant displayed hyposensitivity to cell-wallperturbing agents such as CFW and CR (Fig. 6a, b). Chitin is one of the core polysaccharide components of the fungal cell wall and is critical for maintaining infection-related morphogenesis in phytopathogenic fungi (Geoghegan et al. 2017). The deletion of chitin synthase genes in plant pathogenic fungi results in vegetative developmental defects and reduced virulence (Kong et al. 2012;Liu et al. 2016). In this study, defects in cell wall integrity in the ΔFocGCS mutant were partially caused by the reduced chitin content (Fig. 6c). Our data provide preliminary evidence for the importance of GlcCers in cell wall organization in filamentous fungi. In addition, GCS synthesis inhibition affects membrane trafficking in Giardia lamblia and results in ultrastructural abnormalities including accumulation of cytosolic vesicles, enlarged lysosomes, and clathrin disorganization (Stefanic et al. 2010). As shown in Fig. 6d, we also observed the cytosolic accumulation of vesicles in the ΔFocGCS mutant. The underlying mechanism may be that GlcCer on the cytosolic side of Golgi membrane is necessary for vesicle budding.
To summarize, we characterized FocGCS, a gene that encodes GCS in Foc TR4, and provided several lines of evidence supporting the importance of FocGCS in regulating fungal vegetative growth, conidiation, conidial morphology, and virulence, which also corresponds with changes in the expression level of a series of functional genes in Foc TR4. Furthermore, FocGCS knockout affects cell wall integrity but not tolerance to environmental pH, osmotic, and oxidative stresses. These results establish GlcCer as a key virulence factor of Foc TR4, with important implications for future development of efficient management strategies for Fusarium wilt.

Conclusions
In this study, we explored the biological functions of FocGCS in Foc TR4. Loss of FocGCS resulted in severely retarded fungal vegetative growth, reduced conidia production, substantial loss of virulence, and abnormal conidial morphology. In addition, biochemical analyses showed that FocGCS affects cell wall integrity, but is not essential for tolerance to oxidative and osmotic stresses. Finally, transcriptome analysis showed that the knockout of FocGCS greatly influences transmembrane transport in Foc TR4.

Strains and growth conditions
The Foc TR4 strain II5 (NRRL#54,006), isolated from diseased banana in Guangdong Province of China, was used as the wild-type (WT) strain for transformation, targeted disruption, and complementation experiments in this study. The WT strain and transformants generated in this study were cultivated at 28 °C on PDA (200 g/L potato and 20 g/L dextrose, pH 7.0) medium for mycelial growth and stress tolerance assays. The mung bean liquid (MBL) (20 g mung beans boiled in water for 20 min and brought to 1 L) broth was used for conidiation assay.

Bioinformatics analysis
BLASTp was performed using human GCS protein sequence (GenBank accession no. NP_003349.1) as a query to search for homologs in the NCBI GenBank database. The protein sequences of the GCS protein family were aligned using ClustalX 2.1 and modified by GeneDoc software (Xia et al. 2006). A phylogenetic tree of putative GCS proteins identified in different organisms was constructed through the maximum likelihood (ML) method in MEGA5 software with bootstrap tests. Values on the branches of clusters represent the results of bootstrap analysis (10,000 bootstrap replicates). The Simple Modular Architecture Research Tool (SMART) was used to identify conserved protein domains of FocGCS (Letunic et al. 2021).

Construction of the ΔFocGCS mutant and complementation strains
Targeted disruption of FocGCS was conducted using PEG-mediated protoplast transformation method (Yun et al. 2014). To prepare the FocGCS knockout construct, an 820-bp 5′ flanking fragment and a 971-bp 3′ flanking fragment of FocGCS were amplified with the primer pairs P1/P2 and P3/P4, respectively (Additional file 1: Table S2). The amplified 5′ and 3′ flanking sequences were fused to HPH driven by a constitutive A. nidulans TrpC promoter via double-joint PCR (Liu et al. 2019b).
The fusion construct was transformed into protoplasts of the WT strain II5; the ΔFocGCS mutant was identified through PCR assays with the primers listed in Additional file 1: Table S2 and further confirmed by Southern blot assay. Genomic DNA was isolated from the WT strain and the ΔFocGCS mutant using the Hi-DNA secure Plant Kit (TianGen, Beijing, China) and digested with the XbaI restriction enzyme. Southern blotting was performed using the DIG high prime DNA labeling and detection starter kit II (Roche, Mannheim, Germany) according to the manufacturer's protocol.
To generate the complementation strain, a FocGCS-GFP fusion cassette including the native promoter and coding sequence of FocGCS was inserted into plasmid pYF11, yielding the complementation vector pYF11-FocGCS-GFP. The vector was then transformed into the ΔFocGCS mutant; putative transformants were selected with geneticin and identified by PCR with the gene-specific primers (Additional file 1: Table S2).

Fungal growth and conidial germination assay
Mycelial growth and colony morphology of the WT and mutant strains were measured on PDA after incubation at 28 °C for 5 days. The conidia were harvested from 5-day-old PDA plates and conidial suspension was adjusted to a concentration of 1.0 × 10 6 conidia/mL. For the conidial germination assay, conidial suspension was added into yeast extract peptone dextrose (YEPD: 3 g/L yeast extract, 10 g/L peptone, and 20 g/L, pH 7.0) broth and incubated at 28 °C on a rotary shaker at 180 rpm for 12 h. Conidia were considered to have germinated if the length of the germination tube was more than three times that of the conidia under microscopy. The percentage of conidial germination was determined based on at least 100 randomly selected conidia for each of three replicates.

Fluorescence microscopy analysis
The ΔFocGCS-C-GFP strain was used for fluorescence microscopy analysis of FocGCS expression at different developmental stages. The nuclei of the conidia or hyphae were stained with DAPI (4′,6-diamidino-2-phenylindole, Invitrogen, USA) at a final concentration of 100 nM. For fluorescence microscopic observation of conidial morphology, the conidia were stained with CFW (Calcofluor white M2R, Invitrogen, USA) according to the manufacturer's recommendations. Samples were observed under a confocal laser scanning microscope (LSM 880,Carl Zeiss,Oberkochen,Germany).

Evaluation of stress response
For cell wall sensitivity assays, 5-day-old mycelial plugs were inoculated on PDA plates supplemented with 0.2% CFW (Phytotech, Lenexa, KS, USA) or 0.2% CR (Sigma-Aldrich, St. Louis, MO, USA). The tolerance to multiple stresses was assessed by inoculation of 5-day-old mycelial plugs on PDA containing 1 M NaCl or 1 M sorbitol for osmotic stress or 0.5 mM H 2 O 2 for oxidative stress or with different pH values (4, 6, 8, and 10). After incubation at 28 °C for 5 days, the radial colony growth was measured, and growth inhibition rate was analyzed as previously described (Dai et al. 2016). The inhibition rate (%) = [(the colony diameter of control group -the colony diameter of treated group) / (the colony diameter of control group)] × 100.

Pathogenicity test and fungal biomass estimation
Cavendish banana variety Brazilian (Musa spp. AAA group) plantlets with 6-7 leaves and a height of 30-50 cm were used for pathogenicity test. To prepare the inoculum, 5-day-old mycelial plugs of the tested Foc strains were transferred into MBL broth and grown at 28 °C on a rotary shaker at 150 rpm for 5 days; the spores were collected by filtering through five layers of filter paper and the spore suspension was adjusted to a concentration of 1.0 × 10 7 spores/mL. The trimmed roots of the seedings were dipped into spore suspensions for 2 min and carefully transplanted (Garcia-Bastidas et al. 2019). The disease severity was evaluated at 4 weeks after inoculation using the following rating scales: 0 = no symptom, 1 = little brown spots on the inner rhizome, 2 = less than 25% of the inner rhizome turned brown, 3 = more than 25% and less than 75% of the inner rhizome turned brown, and 4 = dead, full inner rhizome turned dark brown (Liu et al. 2019a).
Banana root samples were collected at 5, 10, and 15 dpi to estimate the fungal biomass inside the plants. The transcript levels of FocEF1α and banana actin gene (ACT1) were determined by qPCR using gene-specific primers (Additional file 1: Table S2). The relative fungal growth in planta was calculated by normalizing FocEF1α to banana ACT1 and shown as ratios of FocEF1α/ACT1 (Dai et al. 2016).
The HPLC-MS/MS analysis was performed on an AB SCIEX 4500 triple quadruple mass spectrometer coupled with AB SCIEX ExionLC AD LC modules (AB SCIEX, Massachusetts, USA). A 5-µL sample was separated on a Phenomenex Luna C8 column (150 mm × 2.0 mm, 3 μm) with a flow rate of 0.3 mL/min at 40 °C. The mobile phase solvent A was 0.2% formic acid and 2 mM ammonium formate in water. The mobile phase solvent B was 0.2% formic acid and 1 mM ammonium formate in acetonitrile. The gradient elution condition of sphingolipids was 80% B at 0-4 min, 80-90% B at 4-10 min, 90-99% B at 10-20 min, and 99% B at 20-26 min, with 80% B at 26-31 min for re-equilibration. Mass instrumentation settings and MRM settings for the targeted sphingolipid species are listed in Additional file 1: Table S4. Processing of the collected data was performed using MultiQuant software (AB SCIEX).

Measurement of chitin content
Chitin content was measured according to a previously described method (Bulik et al. 2003;Dai et al. 2016). The cellular chitin levels were measured using colorimetric determination of β-(1-4)-linked N-acetylglucosamine (GlcNAc). Briefly, 10 mg of ground mycelial powder, from strains cultured in YEPD liquid medium for 3 days, was suspended in 1 mL of 6% KOH and heated at 80 °C for 90 min. After centrifugation at 12,000 g for 10 min, the pellets were suspended in 1 mL phosphate-buffered saline (PBS, pH 7.4) and spun again, and the buffer was discarded. Then, each pellet was suspended in 0.1 mL of McIlvaine buffer (pH 6.0) containing 5 μL chitinase from Streptomyces griseus (5 mg/mL in PBS) and incubated at 37 °C for 24 h. Ten microliters of sample supernatant and 10 μL 0.27 M sodium borate (pH 9.0) were combined and heated for 10 min at 100 °C. After immediately cooling to room temperature, 100 μL of the freshly diluted DMAB solution (10 g of p-dimethylaminobenzaldehyde in 12.5 mL of concentrated HCl and 87.5 mL of glacial acetic acid, diluted 1:10 with glacial acetic acid) was added to the samples and incubated at 37 °C for 20 min. The quantity of glucosamine in samples was calculated by a standard curve prepared with GlcNAc by recording the absorbance at 585 nm.

Transmission electron microscopy
For TEM observation, conidia were collected from 5-day-old PDA plates. Samples were fixed in 2.5% (v/v) glutaraldehyde and 2% (v/v) paraformaldehyde in 0.1 M phosphate-buffered saline (pH 7.4). After being washed with the buffer, specimens were post-fixed for 21 h with 1% potassium permanganate at 4 °C. Samples were embedded in SPI-PON812 resin (SPI Supplies). Ultrathin sections were obtained on a microtome (Leica EM UC6) and stained with 2% uranyl acetate followed by 3% lead citrate. The images were photographed using a transmission electron microscope (JEM-1400Flash) at an accelerating voltage of 120 kV.

Gene expression analysis
Total RNA was isolated from 100 mg of frozen hyphae or banana roots using a Plant RNA Kit (Omega, Georgia, USA). The first-strand cDNA was synthesized from 1 μg total RNA using the PrimeScript 1st Strand cDNA Synthesis Kit (TAKARA, Dalian, China) according to the manufacturer's instructions, and the samples were diluted 1:5 with RNase-free distilled water. qPCR was performed with three technical replicates using the Applied Biosystems StepOnePlus System as described previously (Huang et al. 2019). The FocEF1α gene was used as an internal control to normalize the data for comparing the relative transcript abundance. The relative expression of each gene under different conditions was determined using the 2 −ΔΔCT method (Livak and Schmittgen 2001). All the primer pairs are listed in Additional file 1: Table S2.

Transcriptome analysis
Conidia (1.0 × 10 6 conidia/mL) of the WT and ΔFocGCS strains were cultured in PDB for 3 days at 180 rpm. The hyphae were collected by filtering through five layers of filter paper, washed three times with sterile distilled water, and vigorously ground into a fine powder using liquid nitrogen (Huang et al. 2019). Total RNA was isolated from frozen hyphae powder using the Plant RNA Kit (Omega, Georgia, USA). RNA purity was checked using the NanoDrop spectrophotometer (IMPLEN, CA, USA). The library preparations were sequenced on an Illumina Hiseq 2000 platform.