FgSfl1 and Its Conserved PKA Phosphorylation Sites Are Important for Conidiation, Sexual Reproduction, and Pathogenesis in Fusarium graminearum

The fungal plant pathogen, Fusarium graminearum, contains two genes, FgCPK1 and FgCPK2, encoding the catalytic subunits of cAMP-dependent protein kinase A. FgCPK1 and FgCPK2 are responsible for most of the PKA activities and have overlapping functions in various cellular processes in F. graminearum. The cpk1 cpk2 double mutant was significantly reduced in growth, rarely produced conidia, and was non-pathogenic. In this study, we found that the cpk1 cpk2 double mutant was unstable and produced fast-growing spontaneous sectors that were defective in plant infection. All spontaneous suppressor strains had mutations in FgSFL1, a transcription factor gene orthologous to SFL1 in yeast. Thirteen suppressor strains had non-sense mutations at Q501, three suppressor strains had frameshift mutations at W198, and five suppressor strains had mutations in the HSF binding domain of FgSfl1. Only one suppressor strain had both a non-synonymous mutation at H225 and a non-sense mutation at R490. We generated the SFL1 deletion mutant and found that it produced less than 2% of conidia than that of the wild-type strain PH-1. The sfl1 mutant was significantly reduced in the number of perithecia on carrot agar plates at 7 days post-fertilization (dpf). When incubated for more than 12 days, ascospore cirrhi were observed on the sfl1 mutant perithecia. The infection ability of the sfl1 deletion mutant was also obviously defective. Furthermore, we found that in addition to the S223 and S559 phosphorylation sites, FgSFL1 had another predicted phosphorylation site: T452. Interestingly, the S223 phosphorylation site was responsible for sexual reproduction, and the T452 phosphorylation site was responsible for growth and sexual reproduction. Only the S559 phosphorylation site was found to play an important role in conidiation, sexual reproduction, and infection. Overall, our results indicate that FgSFL1 and its conserved PKA phosphorylation sites are important for vegetative growth, conidiation, sexual reproduction, and pathogenesis in F. graminearum.


Introduction
Fusarium head blight (FHB), caused by Fusarium graminearum, is one of the most important diseases of various cereal crops [1]. The pathogen overwinters on plant debris, and ascospores released from perithecia in the spring are the initial origins of infection [2]. The ascospores are discharged on flowering wheat which become the center of the infection. After the initial colonization, the pathogen spreads from the infection site to other florets and reduces grain quality [2][3][4]. In addition, F. graminearum produces harmful mycotoxins, such as deoxynivalenol (DON) and zearalenone. DON is also phytotoxic and an important virulence factor in the wheat head blight fungus [5][6][7].

Generation of the sfl1 Mutant
For generating the gene replacement construct by the split-marker approach, the 562-bp upstream and 666-bp downstream fragments of SFL1 were amplified with primer pairs FgSFL1/1F-FgSFL1/2R and FgSFL1/3F-FgSFL1/4R (Table S1), respectively. The resulting PCR products were connected to the hph hygromycin phosphotransferase cassette by overlapping PCR and then transformed into protoplasts of PH-1 as described [20]. Hygromycin-resistant transformants were screened by PCR with primers FgSFL1/5F and FgSFL1/6R (Table S1). They were further confirmed by a Southern blot analysis with its downstream flanking sequence as the probe.
For complementation assays, the entire SFL1 gene and its promoter and terminator sequences were amplified with primers FgSFL1/CF and FgSFL1/CR2 (Table S1), digested with SmaI and SacII, and cloned into the NEO R vector pHZ100 [22]. The resulting construct pSFL1 was transformed into protoplasts of the SF-1. The sfl1/SFL1 transformants were verified by PCR.

Plant Infection and DON Production Assays
For infection of wheat cultivar Xiaoyan 22, conidia of PH-1 and mutant strains were harvested from CMC cultures by filtration and re-suspended at 10 5 spores/mL as described [7]. For each flowering wheat head, the fifth spikelet from the base was inoculated with 10 µL of conidial suspensions as described [19,23]. Spikelets with typical symptoms were examined 14 days post-inoculation (dpi) to estimate the disease index [24] and diseased wheat kernels were assayed for DON [22]. For each strain, plant infection and DON production assays were repeated at least three times.

Sexual Reproduction Assays
For self-fertilization, cultures were grown on carrot agar plates [25] for 7 days before pressing down aerial hyphae with a sterile 0.1% Tween-20 solution as described [26]. Perithecium formation, ascus development, and cirrhus production were assayed after incubation at 25 • C under black light for 1 to 2 weeks after fertilization [14,27].

RNA-Seq Analysis
A total of 12 h germlings of the wild-type strain PH-1 and Fgsfl1 mutants were collected to isolate RNA (two biological replicates each). The RNA sample was sequenced with Illumina HiSeq 2500 with the paired-end 2 × 150 bp model at the Novogene Bioinformatics Institute (Beijing, China). All obtained RNA-seq reads (at least 24 Mb each sample) were mapped onto the reference genome of F. graminearum wild-type strain PH-1 [18,29] by HISAT2 [29]. In addition, RNA-seq data were deposited in the NCBI SRA database under the accession number SRR15569465 and SRR15569464. All reads mapped to each gene were calculated by featureCounts [30]. Differentially expressed genes with the false discovery rate (FDR) < 0.05 and log2 fold change (Log2FC > 1) were identified with the edgeRun package as described [31].

qRT-PCR Analysis
A total of 12 h germlings of the wild-type strain PH-1 and Fgsfl1 mutants were collected to isolate RNA (two biological replicates each

The CPK1 CPK2 Mutant Is Unstable and Produces Spontaneous Suppressor Strains with Faster Growth Rate
The cpk1 cpk2 double mutant strain DM-1 was unstable when cultured on PDA at 25 • C. Fast-growing sectors derived from spontaneous mutations often became visible after incubation for 12 days or longer ( Figure 1A). We randomly collected 25 subcultures of spontaneous sectors. Whereas five of them (20%) had a similar growth rate to the wild-type strain PH-1, five (20%) sectors grew faster than the wild-type strain PH-1. All the other suppressor strains grew faster than the original mutant but slightly slower than the wild type ( Figure 1B and Figure S1). Nevertheless, like the cpk1 cpk2 mutant, all the suppressor strains were still defective in conidiation ( Figure 1C). These results indicate that spontaneous mutations in these suppressor strains rescued vegetative growth but did not positively impact conidiation.

Suppressors HS-20 and HS-25 Restored Some Asexual and Sexual Reproduction Defects of the cpk1 cpk2 Double Mutant
We selected suppressor strains HS-20 and HS-25 (Table 1) for further characterization because their growth rate was 98% and 96% of the wild-type strain PH-1, respectively (Table 2; Figure 2A). Although conidium morphology was not affected ( Figure 2B), strains HS-20 and HS-25 were significantly reduced in conidiation ( Figure 1C; Table 2). On self-mating carrot agar plates, the wild-type strain PH-1 produced abundant perithecia and ascospore cirrhi. The cpk1 cpk2 double mutant was sterile and failed to develop perithecia at 7 days post-fertilization (dpf) or longer. Under the same conditions, suppressor strains HS-20 and HS-25 produced fewer perithecia than the wild type ( Figure 2C) and no cirrhi were observed, but they produced normal ascospores ( Figure 2D). Upon continued incubation of HS-20 and HS-25, abnormal white cirrhi were observed from 12-dpf to 14-dpf ( Figure S2A). These results show that the suppressor strains partly restore the sexual reproduction but not the ascospore discharge of the cpk1 cpk2 double mutant. Assay of growth rate of the spontaneous suppressor strains. Mean and standard errors were estimated with data from three independent measurements. (C) Assay of conidiation of the spontaneous suppressor strains. Mean and standard errors were estimated with data from three independent measurements.

Suppressors HS-20 and HS-25 Restored Some Asexual and Sexual Reproduction Defects of the cpk1 cpk2 Double Mutant
We selected suppressor strains HS-20 and HS-25 (Table 1) for further characterization because their growth rate was 98% and 96% of the wild-type strain PH-1, respectively (Table 2; Figure 2A). Although conidium morphology was not affected ( Figure 2B), strains HS-20 and HS-25 were significantly reduced in conidiation ( Figure 1C; Table 2). On selfmating carrot agar plates, the wild-type strain PH-1 produced abundant perithecia and ascospore cirrhi. The cpk1 cpk2 double mutant was sterile and failed to develop perithecia at 7 days post-fertilization (dpf) or longer. Under the same conditions, suppressor strains HS-20 and HS-25 produced fewer perithecia than the wild type ( Figure 2C) and no cirrhi were observed, but they produced normal ascospores ( Figure 2D). Upon continued incubation of HS-20 and HS-25, abnormal white cirrhi were observed from 12-dpf to 14-dpf ( Figure S2A). These results show that the suppressor strains partly restore the sexual reproduction but not the ascospore discharge of the cpk1 cpk2 double mutant.
In infection assays with flowering wheat heads, the wild-type strain PH-1 caused typical head blight symptoms at 14 days post-inoculation (dpi). The cpk1 cpk2 double mutant caused typical symptoms only on the inoculated kernels and failed to spread to neighboring spikelets ( Figure 2E). The suppressor strains HS-20 and HS-25 caused typical symptoms from the vaccination point to neighboring spikelets with a disease index that was 59% and 60.7% that of the wild type ( Table 2). These results suggest that mutations in the suppressor strains HS-20 and HS-25 not only suppressed the defects of the cpk1 cpk2 double mutant in hyphal growth but also in sexual reproduction and plant infection. In infection assays with flowering wheat heads, the wild-type strain PH-1 caused typical head blight symptoms at 14 days post-inoculation (dpi). The cpk1 cpk2 double mutant caused typical symptoms only on the inoculated kernels and failed to spread to neighboring spikelets ( Figure 2E). The suppressor strains HS-20 and HS-25 caused typical symptoms from the vaccination point to neighboring spikelets with a disease index that was 59% and 60.7% that of the wild type (Table 2). These results suggest that mutations in the suppressor strains HS-20 and HS-25 not only suppressed the defects of the cpk1 cpk2 double mutant in hyphal growth but also in sexual reproduction and plant infection.

Identification of Suppressor Mutations in FgSFL1
To identify suppressor mutations, we subjected the three suppressor strains HS-20, HS-21, and HS-25, which had disease indices of 59.0%, 57.4%, and 60.7% of the wild-type index, respectively (Table 2), to the wild-type strain PH-1 and cpk1 cpk2 double mutant for whole genome sequencing (WGS). Interestingly, all three selected suppressor strains had mutations in FgSFL1 (FGRRES_10868), which is orthologous to SFL1 of S. cerevisiae. HS-25 had a non-sense mutation at Q501, and the two others had frameshift mutations at D335 and W198, respectively. We then sequenced the FgSFL1 gene from the other 22 suppressor strains. All of them had mutations in FgSFL1 at ten different sites ( Figure 3). Thirteen suppressor strains had non-sense mutations at Q501, three suppressor strains had frameshift mutations at W198, and five suppressor strains had mutations in the HSF binding domain of FgSfl1 (Table 2). Only one suppressor strain had both a non-synonymous mutation at H225 and a non-sense mutation at R490.

Identification of Suppressor Mutations in FgSFL1
To identify suppressor mutations, we subjected the three suppressor strains HS-20, HS-21, and HS-25, which had disease indices of 59.0%, 57.4%, and 60.7% of the wild-type index, respectively (Table 2), to the wild-type strain PH-1 and cpk1 cpk2 double mutant for whole genome sequencing (WGS). Interestingly, all three selected suppressor strains had mutations in FgSFL1 (FGRRES_10868), which is orthologous to SFL1 of S. cerevisiae. HS-25 had a non-sense mutation at Q501, and the two others had frameshift mutations at D335 and W198, respectively. We then sequenced the FgSFL1 gene from the other 22 suppressor strains. All of them had mutations in FgSFL1 at ten different sites ( Figure 3). Thirteen suppressor strains had non-sense mutations at Q501, three suppressor strains had frameshift mutations at W198, and five suppressor strains had mutations in the HSF binding domain of FgSfl1 (Table 2). Only one suppressor strain had both a non-synonymous mutation at H225 and a non-sense mutation at R490.

Deletion of FgSFL1 Partially Rescued Defects of the cpk1 cpk2 Double Mutant
To verify the suppressor mutations in FgSFL1, the gene replacement construct was transformed into the cpk1 cpk2 double mutant strain DM-1, and we conducted a PCR analysis to confirm that the transformant strains lacked the FgSFL1 gene ( Figure S3). A total of four transformants DF-16, DF-18, DF-23, and DF-25 (Table 1), with the same phenotype were obtained. When the DF-18 was incubated on the PDA medium for 3 d, the colony grew normal and was similar to the suppresser strain HS-25 (Table 3; Figure 4A). Although strain DF-18 was not restored in the reduction of conidiation in the cpk1 cpk2 double mutant (Table 3), the conidium morphology was normal ( Figure 4B).

Deletion of FgSFL1 Partially Rescued Defects of the cpk1 cpk2 Double Mutant
To verify the suppressor mutations in FgSFL1, the gene replacement construct was transformed into the cpk1 cpk2 double mutant strain DM-1, and we conducted a PCR analysis to confirm that the transformant strains lacked the FgSFL1 gene ( Figure S3). A total of four transformants DF-16, DF-18, DF-23, and DF-25 (Table 1), with the same phenotype were obtained. When the DF-18 was incubated on the PDA medium for 3 d, the colony grew normal and was similar to the suppresser strain HS-25 (Table 3; Figure 4A). Although strain DF-18 was not restored in the reduction of conidiation in the cpk1 cpk2 double mutant (Table 3), the conidium morphology was normal ( Figure 4B).
In the sexual stage, the DF-18 produced fewer perithecia with mature ascospores ( Figure 4C,D). However, the cirrhi were not detectable after 7 days post-fertilization (dpf) on the carrot agar plates ( Figure 4C). In the wheat head infection assays, unlike the cpk1 cpk2 double mutant, strain DF-18 caused extensive discoloration beyond the inoculation site at 14 dpi, and the disease index was approximately the same as the suppressor strains ( Figure 4E; Table 3). Therefore, the deletion of FgSFL1 is responsible for all the phenotypes observed in the suppressor strains.   In the sexual stage, the DF-18 produced fewer perithecia with mature ascospores (Figure 4C,D). However, the cirrhi were not detectable after 7 days post-fertilization (dpf) on the carrot agar plates ( Figure 4C). In the wheat head infection assays, unlike the cpk1 cpk2 double mutant, strain DF-18 caused extensive discoloration beyond the inoculation site at 14 dpi, and the disease index was approximately the same as the suppressor strains ( Figure 4E; Table 3). Therefore, the deletion of FgSFL1 is responsible for all the phenotypes observed in the suppressor strains.

FgSFL1 Is Important for Conidiation and Sexual Reproduction
FgSFL1 is predicted to encode a 591-amino acid protein with the typical structure of a heat shock factor (HSF) DNA binding domain ( Figure S4). To determine its function, we generated the sfl1 gene replacement transformant in the wild-type strain PH-1 (Table 1). Five sfl1 mutant strains SF-1, SF-6, SF-39, SF-40, and SF-47 were confirmed by Southern blot hybridization ( Figure S5).
The sfl1 mutant strain SF-1 formed normal colonies similar to the wild-type on the PDA medium ( Figure 5A), but were significantly reduced in conidiation (Table 3). In 5-day-old CMC cultures, the Fgsfl1 deletion mutant produced less than 2% of the conidia than that of the wild-type strain PH-1 ( Table 3). The conidia produced by SF-1 and the Fgsfl1/FgSFL1 transformant CS-1 were normal in appearance, indistinguishable from the wild-type strain ( Figure 5B).

FgSFL1 Is Important for Conidiation and Sexual Reproduction
FgSFL1 is predicted to encode a 591-amino acid protein with the typical structure of a heat shock factor (HSF) DNA binding domain ( Figure S4). To determine its function, we generated the sfl1 gene replacement transformant in the wild-type strain PH-1 (Table 1). Five sfl1 mutant strains SF-1, SF-6, SF-39, SF-40, and SF-47 were confirmed by Southern blot hybridization ( Figure S5).
The sfl1 mutant strain SF-1 formed normal colonies similar to the wild-type on the PDA medium ( Figure 5A), but were significantly reduced in conidiation (Table 3). In 5day-old CMC cultures, the Fgsfl1 deletion mutant produced less than 2% of the conidia than that of the wild-type strain PH-1 ( Table 3). The conidia produced by SF-1 and the Fgsfl1/FgSFL1 transformant CS-1 were normal in appearance, indistinguishable from the wild-type strain ( Figure 5B).
On the self-mating carrot agar plates, the wild-type strain PH-1 and the CS-1 produced abundant perithecia and ascospore cirrhi after 7 dpf ( Figure 5C). Under the same conditions, the Fgsfl1 mutant produced only a few perithecia and no cirrhi ( Figure 5C), and the ascospores did not mature properly ( Figure 5D). After incubation for over 12 days, cirrhi were observed and the perithecia contained mature ascospores ( Figure S2). Thus, losing the function of FgSFL1 leads to a reduced yield of perithecia and delayed sexual development. These results indicate that the FgSFL1 is important for conidiation and sexual development in F. graminearum. On the self-mating carrot agar plates, the wild-type strain PH-1 and the CS-1 produced abundant perithecia and ascospore cirrhi after 7 dpf ( Figure 5C). Under the same conditions, the Fgsfl1 mutant produced only a few perithecia and no cirrhi ( Figure 5C), and the ascospores did not mature properly ( Figure 5D). After incubation for over 12 days, cirrhi were observed and the perithecia contained mature ascospores ( Figure S2). Thus, losing the function of FgSFL1 leads to a reduced yield of perithecia and delayed sexual development. These results indicate that the FgSFL1 is important for conidiation and sexual development in F. graminearum.

FgSFL1 Is Also Important for Infection
In infection assays with flowering wheat heads, the wild-type strain PH-1 caused typical head blight symptoms and had a disease index of 12 ± 1.9 at 14 dpi. The average disease index of the SF-1 was 6 ± 2.2, reduced to almost 50% of the wild-type strain PH-1 ( Figure 5E; Table 3). The CS-1 had a similar disease index as that of the wild-type strain PH-1 ( Figure 5E; Table 3). These results show that the FgSFL1 also plays an important role in plant infection in F. graminearum.

The T452D Mutation in Fgsfl1 Results in Reduced Growth
In S. cerevisiae, Sfl1 has two predicted consensus PKA phosphorylation sites, S207 and S733, that are conserved in FgSfl1 (S223 and S559). In addition, we found that T452 is another putative PKA phosphorylation site because it contains the [S/R] [S/R]-X-[S/T] consensus sequence ( Figure 6) and is conserved in Fgsfl1 and its orthologs from other filamentous ascomycetes ( Figure S4).

FgSFL1 Is Also Important for Infection
In infection assays with flowering wheat heads, the wild-type strain PH-1 caused typical head blight symptoms and had a disease index of 12 ± 1.9 at 14 dpi. The average disease index of the SF-1 was 6 ± 2.2, reduced to almost 50% of the wild-type strain PH-1 ( Figure 5E; Table 3). The CS-1 had a similar disease index as that of the wild-type strain PH-1 ( Figure 5E; Table 3). These results show that the FgSFL1 also plays an important role in plant infection in F. graminearum.

The T452D Mutation in Fgsfl1 Results in Reduced Growth
In S. cerevisiae, Sfl1 has two predicted consensus PKA phosphorylation sites, S207 and S733, that are conserved in FgSfl1 (S223 and S559). In addition, we found that T452 is another putative PKA phosphorylation site because it contains the [S/R] [S/R]-X-[S/T] consensus sequence ( Figure 6) and is conserved in Fgsfl1 and its orthologs from other filamentous ascomycetes ( Figure S4). To determine the role of these three PKA phosphorylation sites, the putative FgSFL1 S223D , FgSFL1 T452D , and FgSFL1 S559D constitutive active alleles were generated and transformed into the Fgsfl1 deletion mutant strain SF-1. As the control, we also generated the FgSFL1 S223A , FgSFL1 T452A , and FgSFL1 S559A alleles and transformed them into SF-1. In the resulting transformants of the Fgsfl1 mutant that were verified for the expression of transforming mutant alleles of FgSFL1, only the Fgsfl1/FgSFL1 T452D transformant P2M-14 grew more slowly than the wild-type strain PH-1 ( Figure 7A). Transformants expressing the other mutant alleles of FgSFL1, including FgSFL1 T452A , were normal in growth ( Figure 7A). To determine the role of these three PKA phosphorylation sites, the putative FgSFL1 S223D , FgSFL1 T452D , and FgSFL1 S559D constitutive active alleles were generated and transformed into the Fgsfl1 deletion mutant strain SF-1. As the control, we also generated the FgSFL1 S223A , FgSFL1 T452A , and FgSFL1 S559A alleles and transformed them into SF-1. In the resulting transformants of the Fgsfl1 mutant that were verified for the expression of transforming mutant alleles of FgSFL1, only the Fgsfl1/FgSFL1 T452D transformant P2M-14 grew more slowly than the wild-type strain PH-1 ( Figure 7A). Transformants expressing the other mutant alleles of FgSFL1, including FgSFL1 T452A , were normal in growth ( Figure 7A). Thus, phosphorylation at T452 negatively affects vegetative growth in F. graminearum. Figure 6. Site-directed mutagenesis of putative PKA phosphorylation sites in FgSfl1. Schematic drawing of the FgSfl1 protein and alignment of the marked region with its orthologs from Neurospora crassa (Nc) and Saccharomyces cerevisiae (Sc). The consensus PKA phosphorylation sites are enclosed with red lines. The putative PKA phosphorylation residues are marked with stars.
To determine the role of these three PKA phosphorylation sites, the putative FgSFL1 S223D , FgSFL1 T452D , and FgSFL1 S559D constitutive active alleles were generated and transformed into the Fgsfl1 deletion mutant strain SF-1. As the control, we also generated the FgSFL1 S223A , FgSFL1 T452A , and FgSFL1 S559A alleles and transformed them into SF-1. In the resulting transformants of the Fgsfl1 mutant that were verified for the expression of transforming mutant alleles of FgSFL1, only the Fgsfl1/FgSFL1 T452D transformant P2M-14 grew more slowly than the wild-type strain PH-1 ( Figure 7A). Transformants expressing the other mutant alleles of FgSFL1, including FgSFL1 T452A , were normal in growth ( Figure 7A). Thus, phosphorylation at T452 negatively affects vegetative growth in F. graminearum.

The S559D Mutation in Fgsfl1 Affects Conidiation and Sexual Reproduction
Our analyses of conidiation indicated that only the Fgsfl1/FgSFL1 S559D transformant P3M-7, similar to SF-1, produced fewer than 4% of the conidia produced by the wild-type strain PH-1 (Table 3). All other strains were not significantly defective in conidiation (Table 3; Figure 7B). These results indicate that the S559 phosphorylation site of the FgSfl1 affects conidiation.
When assayed for sexual reproduction on carrot agar plates, the Fgsfl1/FgSFL1 S559D transformant P3M-7 produced fewer perithecia than the wild-type strain PH-1 but produced more perithecia than the Fgsfl1 mutant ( Figure 7C). After incubation for over 12 days, cirrhi were observed on perithecia ( Figure S2).

Phosphorylation Site in FgSfl1 Essential for Infection
Although the S559 phosphorylation site of FgSfl1 was responsible for conidiation and sexual reproduction and the T452 phosphorylation site of FgSfl1 was responsible for growth, all three of these phosphorylation site mutant strains showed defects in the infection of flowering wheat heads, except the P1A-22 and P3A-7. The average disease indices of the P1M-22, P2M-14, and P3M-7 were 8 ± 1.2, 7 ± 1.5, and 6 ± 2.1, respectively. All the disease indices of the three transformant strains were similar to the SF-1 ( Figure 7E; Table 3). Interestingly, the P1A-22 and P3A-7 had almost no defects in the infection of flowering wheat heads ( Figure 7E; Table 3). The P2A-14 caused typical symptoms only on the inoculated kernels, similar to that of the DM-1. These data indicate that all the phosphorylation sites of FgSfl1 are important for plant infection, and that the T452A and S559A phosphorylation sites of FgSfl1 may affect other processes in F. graminearum.

RNA-Seq Analysis of the Fgsfl1 Mutant
To identify genes affected by the deletion of FgSFL1, we performed an RNA-seq analysis with RNA isolated from hyphae at 5-dpf. In comparison with the wild-type strain PH-1, 263 genes were up-regulated 2-fold or more in the Fgsfl1 deletion mutant (Table S2). Among the lower 49 genes with over 10-fold up-regulation, only one had ortholog to S. cerevisiae (Table S2) fatty acid synthase FAS1, and it was reported that the fatty acid's metabolism is important to the virulence of F. graminearum [32]. The deletion of FgSFL1 appeared to increase the expression of a number of genes related to infectivity, including one hsp70 protein (FGSG_06304), one hsp70 like protein (FGSG_06014), one G-protein coupled receptor protein (FGSG_04923), two glycotransferases, and three ion transporter proteins (Table S2). A gene ontology (GO) enrichment analysis showed that genes encoding membrane, ion transport, or amino acid transport were enriched in genes up-regulated in the Fgsfl1 mutant ( Figure S6). The up-regulation of these genes in the Fgsfl1 mutant suggests that they may be important for infection.
A total of 337 genes were down-regulated in the Fgsfl1 mutant compared to the wild type (Table S3). Of those, 32 were reduced over 10-fold reduction and most of them encoded hypothetical proteins. A GO analysis showed that the down-regulated genes were likely related to an oxidation-reduction process, oxidoreductase activity, transmembrane transport, and membrane ( Figure S7). Many of these down-regulated genes may be related to the development of toxins in the wild type, including FGSG_01000 and FGSG_05839 that are homologous to yeast ERG11 and STL1 [33][34][35]. The deletion of FgSFL1 also reduced the expression of several genes related to toxin synthesis, including the galactose oxidase (FGSG_11032), sarcosine oxidase (FGSG_02894), coenzyme a synthetase (FGSG_06462), and mannose-binding lectin (FGSG_11368). Reduction in the expression of these genes may be partly responsible for toxins produced in the Fgsfl1 mutant.

Discussion
The cAMP-PKA signaling pathway is involved in many cellular progresses in various fungi, such as nutrient sensing, pseudohyphal differentiation, spore germination, and infection processes [10,[36][37][38]. Like many other filamentous ascomycetes, Fusarium graminearum contains two genes, FgCPK1 and FgCPK2, encoding the catalytic subunits of cAMP-dependent protein kinase A. Knocking out both FgCPK1 and FgCPK2 results in severe defects in growth, conidiation, sexual reproduction, and virulence [14]. In addition, the cpk1 cpk2 double mutant was unstable and produced fast-growing spontaneous sectors in the original colony. Suppressor mutations partially recovered some of the defects in the cpk1 cpk2 double mutant. In this study, we chose three suppressor strains with different disease indices for whole-genome sequencing [39] and found that the three selected suppressor strains had a mutation in FgSFL1. The other suppressor strains were detected by a PCR analysis, which indicated that all suppressor strains contained mutations in FgSFL1. In Magnaporthe oryzae, 18 of 20 suppressor strains contained mutations in MoSFL1, which also restores growth defects of the cpkA cpk2 double mutant [17]. Thus, SFL1 is important for regulating cellular processes via the cAMP-PKA signaling pathway.
In Saccharomyces cerevisiae, SFL1 is required for normal cell-surface assembly during vegetative growth, and the deletion of SFL1 enhances pseudohyphal and invasive growth [40,41]. In M. oryzae, although the growth rate of the Mosfl1 mutant was similar to that of the wild-type strain Ku80, conidiation was reduced by approximately 70% [42]. In Candida albicans, the flocculation and filamentation is suppressed by Sfl1, and it acts as a novel negative regulator to control the morphogenesis of C. albicans [43]. We also found that the deletion of SFL1 in F. graminearum negatively affected conidiation but not growth rate. In addition, the sfl1 deletion mutant was defective in perithecia production and reduced by almost 50% in the ability to infect wheat heads, but DON production was not affected (Table S5). This suggests that FgSFL1 does not regulate the expression of the TRI genes [44].
Sfl1 is a downstream target of PKA in yeast and is a substrate of Tpk2 [16]. Its function is controlled by PKA activity [45]. The S207 and S733 were the predicted PKA phosphorylation sites of Sfl1 [46]. Because the tpk1 tpk2 tpk3 triple mutant is non-viable, it is not clear whether the phosphorylation site of Sfl1 could suppress the growth defect of the triple mutant. In M. oryzae, only the MoSFL1 S211D rescued the growth defect of the cpkA cpk2 double mutant but not the appressorium formation [17]. In F. graminearum, the S223 and S559 phosphorylation sites were conserved and aligned with S. cerevisiae ( Figure S4). We also found another potential phosphorylation site, T452, according to the features of the PKA phosphorylation site sequence that was also conserved in M. oryzae, N. crassa, F. verticillioides, and F. oxysporum ( Figure S4). In this study, none of the constitutive active or inactive alleles could imitate the wild-type or Fgsfl1 mutant in all phenotypes that were assayed, indicating that the three phosphorylation sites are precisely regulated by the PKA kinase during different development stages. All the transformant strains had normal growth rates except Fgsfl1/FgSFL1 S452D transformant P2M-22 (Table 3; Figure 7A). In M. oryzae, phosphorylated S211 disrupted its interaction with the Cyc8-Tup1 co-repressor, which prevented its ability to block the transcription of downstream genes that are important for growth [17]. In F. graminearum, the T452 to D mutation may have similar negative effects. In the virulence assay, the Fgsfl1/FgSFL1 S452A transformation strain had a similar disease index to the cpk1 cpk2 double mutant (Table 3), indicating that phosphorylation of T452 by PKA is essential for pathogenesis. During conidial development, only the Fgsfl1/FgSFL1 S559D transformation strain was severely defective in conidiation ( Table 3), suggesting that the non-phosphorylated S559 plays an important role in conidiation. In M. oryzae, the C-terminal region of MoSfl1 interacts with Cyc8 [17]. Thus, the S559 may also affect the interaction with Cyc8 and result in some defects in F. graminearum.
All three transformation strains with non-phosphorylated alleles had a similar sexual reproduction phenotype to the Fgsfl1 mutant. Apparently, all three phosphorylation sites are essential for the function of FgSFL1 in regulating sexual development. However, the S559 to D mutation also was defective in perithecia production and ascospore discharge ( Figure 7C), indicating that, in the complex sexual reproduction process, the S559 phosphorylated state must be precisely controlled during different sexual stages. According to the above results, the functions of the phosphorylation sites are different between F. graminearum and M. oryzae. In F. graminearum, all the three phosphorylation sites are important, as they were responsible for growth, sexual reproduction, and infection progress though the precise regulation of phosphorylation.
All our collected suppressor strains have a mutation in the FgSFL1. In M. oryzae, mutations in MoSFL1 also recovered growth defects of the cpkA cpk2 double mutant [17]. It is likely that SFL1 is prone to mutate in the cpkA cpk2 double mutant, but FgSFL1 was not located in the fast-speed sub-genome with a high frequency of variants [47]. Such a high probability of the mutation in the SFL1 orthologs may indicate that SFL1 plays a conserved role in cellular progresses in filamentous ascomycetes. According to the results of this study, the deletion of FgSFL1 will significantly restore the growth rate of the cpk1 cpk2 double mutant. Therefore, the suppressor strains with a mutation in FgSFL1 are readily detected and recovered. This might obscure more significant mutational events, such as the mutation sites that can fully restore the defects in virulence or sexual development. The cpk1 cpk2 double mutant is severely defective in various processes, such as penetration of wheat epidermis and sexual fruiting body production [14]. The spontaneous suppressor strains obtained during wheat inoculation or on the sexual induction medium may help identify more downstream genes involved in infection and sexual development regulated by the PKA pathway.
In summary, the cpk1 cpk2 mutant was unstable and produced fast-growing sectors. All collected suppressor strains had a colony morphology comparable to the wild-type strain PH-1. They had been identified mutating in FgSFL1 by a whole-genome sequence [39]. By knocking out the FgSFL1 in the cpk1 cpk2 double mutant, we identified the defects of the cpk1 cpk2 double mutant that could be rescued by the loss of FgSFL1. In F. graminearum, SFL1 is important for sexual development, conidiation, and infection. Furthermore, the S559 phosphorylation site is most important in repressing the function of FgSfl1.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/jof7090755/s1, Figure S1: Growth of spontaneous suppressor strains in race tube after cultivation for 15 days, Figure S2: Analysis of ascospores discharge from HS-20, HS-25, DF-18, SF-1, P1A-22, P2A-14, and P3A-7 on carrot plate for 12 days, Figure S3: Putative sfl1 cpkA cpk2 mutants were screened by PCR analysis, Figure S4: Sequence alignment of SFL1, Figure S5: Putative sfl1 mutants were screened by southern blot analysis, Figure S6: GO analysis of up-regulated genes in the 12-hours hyphae of the wild-type strain PH-1 and the Fgsfl1 mutant, Figure S7: GO analysis of down-regulated genes in the 12-hours hyphae of the wild-type strain PH-1 and the Fgsfl1 mutant, Figure S8: qRT-PCR analysis of the wild-type stain PH-1 and the Fgsfl1 mutant, Table S1: PCR primers used in this study, Table S2: Genes up-regulated in the Fgsfl1 mutant in comparison with the wild type, Table S3: Genes down-regulated in the Fgsfl1 mutant in comparison with the wild type, Table  S4: Analysis DON production in the Fgsfl1 mutant.