The fungal strains used in this study include wild-type strains F. graminearum NRRL 46422 and PH-1. Cultures were maintained on V8 agar with a 12:12 h light/dark cycle at 28°C with ultraviolet light. Spring wheat, the susceptible Norm cultivar, was grown in controlled growth chambers under a 16:8 h light/dark cycle at 23:20°C with 50% humidity.

The amino acid sequence of FgShyC (GenBank Accession No. BK010685), which was identified from our previous study, was used to search for SA hydroxylase homologs in the genome of F. graminearum PH-1 by Basic local alignment search tools (pBLAST). Eight salicylate hydroxylase candidates, FGSG_03657 (FgShy1), FGSG_09063, FGSG_00092, FGSG_10612, FGSG_05063, FGSG_04776, FGSG_10643 and FGSG_08116, were identified. Shy1 from U. maydis (Accession No. XM_756284) and NahG from P. putida (Accession No. J05317) were also included in the analysis. Sequence alignment was performed and a phylogenic tree was constructed using the MEGA7 program (Kumar et al., 2016) with the Maximum likelihood method (Whelan and Goldman, 2001) using a bootstrap value of 1000.

Fusarium graminearum NA1 strain PH-1, NA2 strain 46422, FgshyC and Fgshy1 mutants were tested for the ability to degrade SA in mung bean liquid medium. Approximately 2 × 10⁵ conidia were inoculated into 4 mL mung bean liquid medium in a 50 ml tube (Cell Treat Scientific Products, Pepperell, MA, United States). Cultures were grown in an unlit shaker (200 rpm, 28°C) for 24 h. SA was dissolved in 2% methanol with the final concentration at 100 mM and was adjusted to pH 8.0. Then 1 mM SA was added into the culture, which was grown for another 24 h in the same condition. Sterile mung bean liquid medium containing 1 mM SA in mung bean liquid medium was included for relative quantification. Since it was determined that fungal growth in mung bean medium for 24 h significantly increased the pH from 6.5 to approximately 8.3, all samples were adjusted to pH around 3.0 before SA extraction.

Salicylic acid was extracted from the 4 mL cultures by adding an equal volume of HPLC grade methylene chloride (Sigma-Aldrich, St. Louis, MO, United States). As an internal standard to normalize sample extraction efficiencies, 0.1 mM of deuterated SA (3, 4, 5, 6-D4, 97%) (Cambridge Isotope Laboratories, Inc., Andover, MA, United States) was added to each sample and immediately mixed on a vortex fitted with a horizontal 15 mL tube holder for 2 min at max speed. The two phases were then separated by centrifugation at 4500 rpm for 3 min. One mL of the organic phase was transferred to a 4 mL glass vial and derivatized using 5 μL trimethylsilyldiazomethane (Sigma-Aldrich, St. Louis, MO, United States) for 30 min. Derivatization converted SA into the more volatile compounds, methyl salicylate (SA-ME) and trimethylsilyl methyl salicylate (SA-TMS), which were then collected by vapor-phase extraction and analyzed by gas chromatography/chemical ionization-mass spectrometry (GC/CI-MS) using a previously reported method (Schmelz et al., 2004; Vaughan et al., 2014). The peak areas of SA-ME and SA-TMS from each sample were summed, divided by the peak area of the deuterated SA, and then divided by area estimated from extractions of the average peak area of samples from sterile media containing 1 mM SA. Three biological replicates were analyzed for each treatment. Differences between means were determined using the Dunnett’s test which compared all other means to the sterile media containing 1 mM SA. All statistical analyses were performed with JMP statistical software (V13.1.0).

RNA was extracted from wild-type 46422 supplemented with 1 mM SA or methanol only as control. First-strand cDNA was synthesized and reverse transcription quantitative polymerase chain reaction (RT-qPCR) was run with Bio-Rad real time polymerase chain reaction (PCR) machine. Gene expression levels were compared with the 2^-ΔΔct values using Bio-Rad CFX manager. Gene specific primers from FgShyC, FgShy1 and salicylate hydroxylase candidates (Supplementary Table S1) were used for transcripts quantification and β-tubulin was amplified and used as internal control for gene expression normalization. For each gene, the transcripts from the methanol were set as one for calculation.

Heterologous E. coli strains were constructed to express FgShyC and FgShy1. The latter is identical to the theoretical protein described by NCBI reference sequence XP_011322042.1. P. putida NahG (You et al., 1991; UniProtKB P23262.4) was used as a positive control, as described previously (Rabe et al., 2013). Commercial expression plasmid pRSET A (Thermo Fisher) was cleaved by NdeI and HindIII restriction endonucleases. The two resulting DNAs were separated by agarose gel electrophoresis and the larger was purified. Two genes were synthesized, each encoding predicted SA hydroxylase proteins with amino-terminal GST fusions (Integrated DNA Technologies, Coralville, IA, United States). Each synthetic gene was cloned into the purified plasmid DNA by the method of Gibson (Gibson et al., 2009), using a commercial kit (New England Biolabs, Ipswich, MA, United States). Plasmid clones were verified by agarose gel analysis and DNA sequencing. Each plasmid was transformed into E. coli BL21 (DE3) pLYSs for expression studies and agar plate assays.

Expression studies were initiated by inoculating Luria-Bertani liquid medium (100 mL) from stationary phase cultures (1 mL inoculum). Cultures were grown in a shake incubator at 37°C for approximately 2 h (OD₆₀₀ = 0.35). Protein expression was induced by addition of IPTG (1 mM final) and continued for 2 h. After expression, cells were precipitated from cultures by centrifugation and suspended in 0.6 mL lysis buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl). Suspended cells were placed in 1.5 mL tubes containing 300 mg of beads (0.5 mm diameter zirconium oxide beads) and lysed in a bullet blender (Next Advance, Troy, NY, United States). Samples were analyzed by SDS-PAGE followed by whole protein staining with Coomassie stain.

Agar plate assays were performed to determine if the expressed proteins were SA hydroxylases. Expression strains were streaked from mature liquid cultures onto plates containing Luria-Bertani medium with IPTG and SA (10 g/L tryptone, 5 g/L yeast extract, 5 g/L NaCl, 15 g/L agar, 0.01 mM IPTG, 1 mM SA). Plates were incubated at 25°C and monitored for development of brown color, indicating conversion of SA to catechol, followed by its oxidation. The strain expressing F. graminearum FgShy1 was compared to both a positive control expressing NahG and a negative control expressing bacterial beta-galactosidase.

OSCAR and protoplast mediated transformation was used to create FgshyC and Fgshy1 mutants following published protocols (Paz et al., 2011). A DNA fragment upstream of the start codon and a fragment downstream of the stop codon of FgShyC or FgShy1 were PCR amplified from 46422 and PH-1 genomic DNA respectively with primer pairs including the BP cloning site (Supplementary Table S1). BP reactions were performed with OSCAR plasmids and products were transformed into E. coli competent cells. Correct constructs were confirmed by PCR analysis using two primers pairs, OSC-F and Hyg-R210, and Hyg-F850 and OSC-R (Supplementary Table S1). Then these constructs were used to obtain the fragment containing hygromycin gene (Hyg) and upstream and downstream of target gene for protoplast transformation. Hygromycin resistant isolates resulting from transformation were transferred to new V8 juice plates containing hygromycin for DNA isolation. Replacement of FgShyC or FgShy1 with the hyg fragment was confirmed by PCR. Whole genome sequencing was performed to confirm the replacement of target gene with Hyg and the absence of ectopic Hyg insertion (Kelly and Ward, 2018).

The effect of SA on the growth of the Fgshy1 mutant was tested on 1% agar plates containing SA. Wild-type strain PH-1 served as control. Agar plates were supplemented with 0.1, 0.5, 1, or 2 mM SA from a 1 M SA stock in methanol. Plates with added methanol were used as controls. Each plate was inoculated with a plug from V8 plate. The plates were incubated at room temperature in the dark. The radial growth of mycelia was measured on day 5. Four biological replicates were performed.

To test the growth of PH-1 and Fgshy1 mutants, 10 mM SA was added to plates and the pH was raised to 8.0. The radial growth of mycelia for PH-1 and Fshy1 mutant was measured and compared. Four biological replicates were performed. Comparisons between mean radial growth measurements were made using an ANOVA.

Fusarium graminearum typically produces 15-ADON or 3-ADON in agmatine liquid culture. 15-ADON was measured in PH-1, Fgshy1 mutants M1, M2, and M18 grown in liquid agmatine media (Gardiner et al., 2009). Cultures were grown in 20 mL of agmatine media for 7 days on a rotary shaker at 200 rpm in the dark at 28°C and then extracted with 8 mL of ethyl acetate. GC–MS analysis was performed on a Hewlett Packard 6890 gas chromatograph fitted with a HP-5MS column (30 m, 0.25 mm, 0.25 μm) and a 5973 mass detector.

For gene expression studies, wheat heads were inoculated with F. graminearum by dipping the whole wheat head into a 10⁵ conidia/ml suspension containing 0.02% Tween 20. Experiments were performed in triplicate. Wheat heads were collected at 3, 6, 24, and 48 h post-inoculation. Two heads were collected and combined for each time point. Three replicates were performed. Wheat heads were lyophilized and pulverized as described above. RNA was extracted from pulverized wheat head tissue using Trizol combined with column purification and on column digestion according to the manufacture’s instruction (Thermo Fisher Scientific, Waltham, MA, United States). cDNA synthesis, qPCR and gene expression analysis were performed as described above.

Fusarium graminearum PH-1 and Fgshy1-M1 and M2 mutant inocula were prepared from 4-day old mung bean cultures. The macroconidia were passed through a 45 μM strainer and adjusted to a concentration of 1 × 10⁵ conidia/mL for point inoculation. Twenty four heads from four pots were used for each strain. Wheat heads at flowing stage were inoculated with 10 μL of F. graminearum PH-1 or Fgshy1 mutants. Inoculated wheat heads were covered with plastic bag for 3 days to keep high humidity. FHB disease development was scored at 7, 10, 14, 17, and 21 days post-inoculation (dpi). Disease severity was measured as a percentage of florets displaying symptoms. The mean percent disease for each treatment was compared independently for each time point using an ANOVA.

During the comparative analysis of genome sequences from 60 strains representing three North American F. graminearum populations, we identified a salicylate hydroxylase homolog (designated FgShyC, for salicylate hydroxylase candidate) unique to the NA2 (North American population 2) (Kelly and Ward, 2018). FgShyC contains 428 amino acids (aa), and shares the highest (81%) identity to a salicylate hydroxylase homolog from F. proliferatum, and 35% identity to the well-studied NahG from P. putida (You et al., 1991). Sequence comparisons and homolog search results showed that FgShyC is present only in NA2 strains, which are more aggressive than the NA1 population (Ward et al., 2008; Foroud et al., 2012). Therefore, we hypothesized that FgShyC provides the NA2 population a pathogenic advantage by degrading SA and suppressing the SA-mediated defense response of wheat.

A previous study using high performance liquid chromatography (HPLC) analysis showed that F. graminearum strain DAOM180378 can metabolize SA (Qi et al., 2012). We examined the ability of F. graminearum strains PH-1 and NRRL 46422 to degrade SA in mung bean liquid medium. GC/MS analysis showed that only a trace of amount of SA was detected in PH-1 and 46422 cultures 24 h after addition of 1 mM SA, whereas the SA level was not altered in media controls without fungal inoculation (Figure 1). This demonstrated the ability of PH-1 and 46422 to degrade SA.

To examine whether FgShyC is involved in SA degradation, FgshyC of F. graminearum NRRL 46422 was disrupted using the split marker approach. The FgshyC mutant was identified by screening over sixty hygromycin resistant transformants by PCR amplification. The replacement of FgShyC was confirmed by genome sequencing and analysis. The ability of wild-type 46422 and the FgshyC mutant to metabolize SA was tested. When SA was added to liquid cultures of either wild-type 46422 or the FgShyC mutant it was degraded similarly (Figure 1). This demonstrates that F. graminearum produces salicylate hydroxylases and argues against this role for FgShyC. In addition, recombinant FgShyC protein expressed in E. coli (Figure 4A) did not convert SA to catechol in a colorimetric plate assay (not shown). Taken together, these results indicate that although FgShyC had sequence homology with salicylic hydroxylase genes, it did not function as a salicylate hydroxylase in our assays, and indicate that at least one other salicylate hydroxylase is active and responsible for the SA degradation activity observed in F. graminearum.

To identify additional salicylate hydroxylase candidates, we searched for FgShyC homologs in the F. graminearum PH1 genome by BLAST. Sequence examination and analysis identified eight salicylate hydroxylase homologs in the PH-1 genome. These eight candidates, which are also present in the NA2 strain 46422, share about 35% identity with each other and Shy1 from U. maydis and NahG from P. putida (You et al., 1991; Rabe et al., 2013). The salicylate hydroxylases from bacteria and fungi are flavin-dependent monooxygenases that contain the conserved FAD and NADH binding sites. These domains were identified in the eight salicylate hydroxylase homologs of F. graminearum except that FAD1 fingerprint 1 is absent in FGSG_10643 (Supplementary Figure S1). A phylogenetic analysis of these salicylate hydroxylase homologs was performed (Figure 2). FGSG_03657 (designated FgShy1) is most closely related to NahG from P. putida and Shy1 from U. Maydis, which degrades SA (You et al., 1991; Rabe et al., 2013).

Real-time PCR was used to determine if SA induces expression of any of the salicylate hydroxylase candidates. F. graminearum 46422 was grown in liquid cultures, with and without SA, and RNA was isolated. Comparative expression results revealed that one candidate, FgShy1, was highly induced by SA (Figure 3). FgShy1 transcripts were upregulated about 400-fold 2 h after addition of SA. FGSG_09063 was also induced, about 30-fold, whereas the other six salicylate hydroxylase homologs including FgShyC were not induced. These results demonstrate that FgShy1 is highly induced by SA and suggest that it may be an active salicylate hydroxylase.

The salicylate hydroxylase activity of FgShy1 was confirmed by creating a heterologous strain of E. coli that produces recombinant FgShy1 from the encoding sequence and testing its ability to degrade SA using an in vivo plate assay. P. putida NahG (You et al., 1991) was used as a positive control. Both FgShy1 and NahG were expressed as fusions to Glutathione S-transferase as this was previously reported to improve NahG solubility (Rabe et al., 2013). Expression of both recombinant proteins was confirmed by inducing liquid cultures and analyzing their cellular protein content by SDS-PAGE (Figure 4A). Both strains produced large amounts of protein similar to the predicted sizes of 73.6 kDA (NahG) and 74.3 kDA (FgShy1) for the GST fusions. When grown on agar plates containing SA both the control NahG and FgShy1 producing strains turned the plates brown, as a result of auto-oxidation of catechol, and indicative of SA conversion to catechol (Figure 4B).

To determine if FgShy1 is essential for SA degradation in F. graminearum, deletion mutants were created by replacing FgShy1 with a hygromycin-resistance gene (Hyg). A total of 19 hygromycin resistant transformants were obtained. PCR amplification indicated that three of the transformants lacked the FgShy1 coding region (Supplementary Figure S2). All three Fgshy1 mutants (M1, M2, and M18) grew the same as wild-type on plates and in liquid media. FgShy1-M1, M2, and M18 were examined for their ability to degrade SA in mung bean culture. GC/MS analysis showed that the wild-type strain PH-1 degraded approximately 80% of the added SA whereas Fgshy1 deletion mutants were significantly reduced in their ability to degrade SA in liquid culture. Compared to media control, Fgshy1-M1 and M2 mutants were able to degrade 10 and 11% of the SA respectively but Fgshy1-M18 did not significantly reduced the amount SA added (Figure 5). Therefore, Fgshy1 is the key active enzyme involved in the observed SA degradation by F. graminearum, but there may be at least one other protein that modestly contributes to the activity found in the wild-type strain.

Previous work demonstrated that F. graminearum growth is inhibited by SA under acidic growth conditions whereas F. graminearum can use SA as a carbon source under basic conditions (Qi et al., 2012). We therefore examined whether or not SA affects growth of the Fgshy1 mutant differently than the wild-type PH-1 strain under acidic and neutral conditions. As expected, with an increase in SA concentration, the colony growth of both Fgshy1 and PH-1 was significantly reduced under acidic conditions. With addition of SA at concentrations of 1 and 2 mM, the Fgshy1 mutant grew slightly slower than the wild-type PH-1 and the hyphae turned yellow (Figure 6A). However, no significant growth difference between Fgshy1 mutant and PH-1 was found by statistical analysis (Figure 6B). Additionally, growth of the Fgshy1 mutant and PH-1 were not inhibited when grown on basic agar media (pH 8.0) containing 10 mM SA (Figure 6C). Our results suggest that FgShy1 is not essential for the growth of F. graminearum on agar medium with SA, suggesting additional enzymes or other SA degradation pathways exist.

To determine whether FgShy1 and its homologs are induced during wheat head infection, we examined their transcript levels at various time points after dip inoculation with F. graminearum 46422. FgShy1 expression was induced about fourfold at 3 hpi Similarly, FGSG_09063 was induced about threefold at 3 hpi and FGSG_10643 was upregulated about 2- and 6-fold at 24 and 48 hpi respectively (Figure 7). The remaining six FgShy1 homologs did not show signs of induction (data not shown). Induction of FgShy1, FGSG_09063, and FGSG_10643 in infected wheat heads suggests they may play a role in FHB development.

A previous study showed that DON production in F. graminearum is reduced by addition of SA (Qi et al., 2012). We measured DON production in Fgshy1 mutant in agmatine liquid medium. Our results demonstrated that 15-ADON production was similar in the parent strain PH-1 and the Fgshy1 deletion mutants (not shown). However, the induction of FgShy1 in infected wheat raises the possibility that FgShy1 degrades plant-produced SA and suppresses SA-mediated defense response. To test if disruption of FgShy1 affect FHB development, we performed virulence assays on wheat heads using Fgshy1 mutants M1 and M2. Head blight development was assessed following point inoculation with Fgshy1 mutants and the wild-type strain PH-1 over a 21-day period. There was no significant difference between the wheat heads inoculated with Fgshy1 mutants and those inoculated with PH-1 (Figure 8). This suggests that disruption of FgShy1 alone has no significant effect on F. graminearum pathogenesis.