Commercial hybrids and mutant genotypes reveal complex protective roles for inducible terpenoid defenses in maize

Abstract

Plant defense research is facilitated by the use of genome-sequenced inbred lines; however, a foundational knowledge of interactions in commercial hybrids remains relevant to understanding mechanisms present in crops. Using an array of commercial maize hybrids, we quantified the accumulation patterns of defense-related metabolites and phytohormones in tissues challenged with diverse fungal pathogens. Across hybrids, Southern leaf blight (Cochliobolus heterostrophus) strongly elicited specific sesqui- and diterpenoid defenses, namely zealexin A4 (ZA4) and kauralexin diacids, compared with the stalk-rotting agents Fusarium graminearum and Colletotrichum graminicola. With respect to biological activity, ZA4 and kauralexin diacids demonstrated potent antimicrobial action against F. graminearum. Unexpectedly, ZA4 displayed an opposite effect on C. graminicola by promoting growth. Overall, a negative correlation was observed between total analyzed terpenoids and fungal growth. Statistical analyses highlighted kauralexin A3 and abscisic acid as metabolites most associated with fungal suppression. As an empirical test, mutants of the ent-copalyl diphosphate synthase Anther ear 2 (An2) lacking kauralexin biosynthetic capacity displayed increased susceptibility to C. heterostrophus and Fusarium verticillioides. Our results highlight a widely occurring defensive function of acidic terpenoids in commercial hybrids and the complex nature of elicited pathway products that display selective activities on fungal pathogen species.

Fungal pathogens, kauralexins, maize defense, phytoalexins, phytohormones, zealexins

Introduction

Crop yield losses due to biotic and abiotic stress occur throughout the world, leading to billions of dollars in lost revenue annually (Oerke, 1994; Savary et al., 2012). Significant contributing factors are pathogenic fungi that attack both below- and above-ground tissues, including the contamination of seeds with carcinogenic mycotoxins (Chassy, 2010). Inducible protective responses to these pathogens are often mediated by defense signaling molecules including reactive oxygen species, jasmonates, ethylene, and salicylates (Glazebrook, 2005). Abscisic acid (ABA) and indole-3-acetic acid (IAA) are plant hormones primarily known for their roles in drought stress and development, respectively, but also have selective roles in plant defense (Jia et al., 2001; Tiryaki and Staswick, 2002; Adie et al., 2007; Erb et al., 2009; Kazan and Manners, 2009; Ton et al., 2009; Vaughan et al., 2015). The collective role of many phytohormone signals is the modulation of inducible gene expression, the control of diverse proteins, and regulation of the biosynthesis of specialized metabolites that directly counter microbial invaders.

A dramatic, rapid, and local maize defense response predicted to directly impede pathogen spread is the inducible production of non-volatile acidic terpenoids that accumulate after exposure to microorganisms or abiotic stress (Schmelz et al., 2014; Vaughan et al., 2015). The existence of locally produced antibiotics, known as phytoalexins, was recognized >70 years ago during the examined interaction of potato tubers with Phytophthora infestans (Müller and Börger, 1940). Monocot phytoalexin research was later initiated in the 1970s following the identification of inducible diterpenoids in rice and subsequent classification into structurally distinct groups including phytocassanes A–E (Koga et al., 1995, 1997; Yajima and Mori, 2000), oryzalexins A–F (Akatsuka et al., 1983, 1985; Kono et al., 1984, 1985), momilactones A and B (Cartwright et al., 1977, 1981), oryzalexin S (Okada, 2011; Shimizu et al., 2008), and ent-10-oxodepressin (Inoue et al., 2013). Analyses of diterpenoid defenses have not yet been well defined in rice, as contributions to both resistance (Peters, 2006; Toyomasu, 2008; Hasegawa et al., 2010; Toyomasu et al., 2014) and susceptibility (Xu et al., 2012) to Magnaporthe oryzae have been observed.

In contrast to rice, diterpenoid defenses in maize have only been described relatively recently (Schmelz et al., 2011). First detected in stem tissues attacked by the European corn borer larvae (ECB; Ostrinia nubilalis), six acidic ent-kaurane-related diterpenoids, termed kauralexins, were initially identified. Based on NMR-elucidated structural similarities, mass spectra, and co-regulation, the saturated and monounsaturated diterpene acids were assigned as kauralexin A1 (KA1) and kauralexin B1 (KB1), respectively, and their logical derivatives as KA2, KA3 and KB2, KB3 (Bohlmann et al., 1982; Ellmauerer et al., 1987; Schmelz et al., 2011). As an initial estimate of antimicrobial activity, liquid culture assays demonstrated growth-inhibitory activity for KA3 and KB3 against fungi commonly isolated from maize, including Rhizopus microsporus and Colletotrichum graminicola. Strongly elicited by multiple fungal pathogens, the accumulation of kauralexins is preceded by the induction of transcripts encoding the ent-copalyl diphosphate synthase known as Anther Ear 2 (An2), an ortholog of rice ent-copalyl diphosphate synthase genes that produce diterpenoid defenses (Harris et al., 2005; Watanabe et al., 1996). In support of the endogenous role of An2, a transposon insertion resulting in an an2 null mutation results in the elimination of maize kauralexin production (Vaughan et al., 2015).

Additional early insights into the complexity of antifungal terpenoid defenses in maize were provided by the genetic and biochemical analyses of the nearly identical sesquiterpene synthases 6 and 11 (Tps6/11) that cyclize farnesyldiphosphate to produce (S)-β-bisabolene and ultimately (S)-β-macrocarpene (Basse, 2005; Kollner et al., 2008). Consistent with Tps6/11 genes being among the most highly up-regulated transcripts in response to Fusarium graminearum, 14 novel analytes were detected in F. graminearum-infected stem tissues (Huffaker et al., 2011). Purification and structural elucidation efforts have resulted in the identification of five abundant acidic sequiterpenoids, termed zealexins (Huffaker et al., 2011; Christensen et al., 2017). Derived from β-macrocarpene, zealexin A1 (ZA1) can be enzymatically produced in vitro by a maize cytochrome P450 (Mao et al., 2016), which undergoes additional hydroxylations at the C1 and C8 positions to produce ZA2 and ZA3, respectively (Huffaker et al., 2011). Synthesis of ZA4 probably occurs via additional hydroxylation activity on ZA3, with the transformation of the C8 alcohol into a geminal diol that spontaneously dehydrates into a ketone (Greer et al., 2007). A desaturation event on ZA1 results in an additional double bond at C1–C6 to yield ZB1 (Huffaker et al., 2011). The capacity for zealexin production appears to be ubiquitous in diverse maize lines, broadly inducible by fungal pathogens and to a lesser extent by stem-boring insects. In liquid culture bioassays, zealexins exhibit potent antibiotic activity against R. microsporus, Aspergillus flavus, and F. graminearum, supporting a direct defensive role against fungal proliferation (Huffaker et al., 2011; Christensen et al., 2017).

In our current study we sought to quantify patterns of elicitation, abundance, and diversity of kauralexins and zealexins in multiple commercial lines that have contributed to large-scale maize production and better represent agriculturally relevant field plantings. Towards this goal, we profiled zealexins, kauralexins, and defense hormones in nine commercial hybrid lines (CHLs) infected with four pathogenic fungi, namely Cochliobolus heterostrophus, C. graminicola, Fusarium verticillioides, and F. graminearum. Distinct defense accumulation patterns were detected for acidic terpenoids in pathogen-infected CHLs, and an overall negative correlation between phytoalexin production and fungal growth was observed. Interestingly, a comparatively modest elicitation of ZA4, KA2, and KB2 occurred in response to F. graminearum; however, ZA4, KA2, and KB2 exhibited strong antimicrobial activity against F. graminearum in liquid cultures. The result suggests that F. graminearum selectively benefits from suppressed levels of specific elicited metabolites. Among the defense metabolites measured in infected CHL stems, the predominant diterpenoid KA3 had the strongest statistical impact associated with reduced fungal growth. Additionally, the drought-inducible phytohormone ABA was the second most statistically influential metabolite correlated with mediating fungal growth. On examining the biological significance of KA3 and related metabolites more closely, an2 mutants lacking kauralexins were found to display significantly greater C. heterostrophus and F. verticillioides growth than the respective wild-type plants. Collectively, these results affirm a prominent yet complex role for maize terpenoid networks mediating defense against pathogenic fungi.

Materials and methods

Plant and fungal material

Seeds from nine CHLs, namely B73xMo17, and the Monsanto lines DK687, RX813, DK580, DK657, DK415, DK435, and DK604, were germinated in MetroMix 200 (SunGro Horticulture Distripution, Inc., Bellevue, WA, USA) and grown as previously described by Schmelz et al. (2011). For analysis of ABA-deficient vp14 mutants in the B73 background, plants were grown in an eight-room sunlit greenhouse on the University of Florida campus at Gainesville as previously described by Vaughan et al. (2015). Plastic pots (54 mm×54 mm×63.5 mm) filled with MetroMix 200 (Sun Gro Horticulture Distribution, Inc.) were used for plant growth. Temperatures were maintained at 28 °C from 07.00 h to 19.00 h Eastern Standard Time followed by a 1 h transition to 22 °C overnight. Relative humidity was controlled between 55% and 60%. For stem and kernel inoculations, single isolates of C. graminicola, F. graminearum, C. heterostrophus, or F. verticillioides were cultured at 28 °C day/22 °C night on a 12 h/12 h light/dark regime for 2 or 3 weeks on V8 agar or, for C. graminicola, on oatmeal agar for 1–3 weeks, prior to experimental use of spores. While C. graminicola and F. graminearum naturally colonize maize stem tissue, it is less common for C. heterostrophus stem infections to occur naturally (e.g. the colonization of stems of cytoplasm male-sterile Texas lines with C. heterostrophus race T). However, to be consistent in defense metabolite comparisons of CHL responses to diverse fungi, inoculations with C. heterostrophus were also carried out in the stems.

Maize stem and kernel metabolite elicitation assays

Fungal and chemical elicitation assays in slit maize stems were performed on 28- to 32-day-old CHLs (B73+Mo17, DK687, RX813, DK580, DK657, DK415, DK435, and DK604) or wild-type (WT) and an2 mutants in the W22 background as previously described (Schmelz et al., 2011). Briefly, plants in damage-related treatment groups received an 8–10 cm parallel longitudinal incision using a surgical scalpel through the center of the upper nodes, internodes, and unexpanded leaves. Infected plants were equally incised and 100 µl of fungal spore suspension (1 × 10⁶) was pipetted evenly over the wound site. For vp14 mutant treatments, experiments were performed on 32-day-old B73 (WT) and vp14 mutants. Inoculations were performed as previously described (Christensen et al., 2015). In brief, 32-day-old plants were injected 25 times with a 22 gauge needle through the developing midribs every 0.5–1 cm spanning the apical node to the base of the whorl, dispensing ~0.3 µl of C. heterostrophus spore suspension per injection. Damage controls were similarly pierced in the same fashion with a 22 gauge needle without spore suspension. Forty-eight hours post-inoculation, inner whorl stem tissue surrounding the treatment site was harvested in liquid nitrogen. For W22 and an2 mutant kernel inoculation with F. verticillioides, methods were followed as described by Christensen et al. (2012) and samples were harvested in liquid N₂ 5 days post-inoculation for metabolite analysis.

Quantification of maize metabolites

For salicylic acid (SA), cinnamic acid (CA), jasmonic acid (JA), IAA, ABA, 12-oxo-phytodienoic acid (12-OPDA), kauralexin, and zealexin quantification, samples were solvent extracted, methylated, collected on a polymeric adsorbent using vapor phase extraction (VPE), and analyzed using GC/isobutene chemical ion MS (CI-MS) as previously described (Schmelz et al., 2011). Metabolite quantification was based on d₆-SA (Sigma-Aldrich, St. Louis, MO, USA), d₅-JA and d₅-CA (C/D/N Isotopes Inc, Pointe-Claire, Canada), or U-[¹³C]18:3 (Cambridge Isotope Laboratories, Inc., Tewksbury, MA, USA) as internal standards. Analysis of ergosterol was carried out as described by Dong et al. (2006) using MeCl₂ extraction as described by Schmelz et al. (2004). Following MeCl₂ extraction of 100 mg of tissue with 10 µl of 7-dehydro-cholesterol (1 mg ml^–1) as internal standard, the organic phase was transferred to a 4 ml vial, and dried under a nitrogen stream. Next, 200 µl of hexane and 1 ml of base (5% KOH in 7:3 MeOH:H₂O) was added then samples were vortexed for 10 s and incubated for 1 h at 80 °C. Following incubation, 1 ml of H₂O and 2 ml of hexane were added and samples were vortexed for ~10 s. The hexane layer was then transferred to a 4 ml vial and dried under a nitrogen stream. Metabolites were resolubulized in 50 µl of chloroform, transferred to a glass GC insert, dried down under nitrogen, and derivatized with 20 µl of N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA) at 80 °C for 1 h. Following incubation, 150 µl of dry MeCl₂ was added and samples were analyzed by GC/CI-MS.

Anti-fungal activity assays

Fungal anti-growth activity assays were modified from the Clinical and Laboratory Standards Institute M38-A2 guidelines as described by Schmelz et al. (2011). The spore suspension (5 × 10⁴ spores ml^–1) was combined with nutrient medium containing the test compound dissolved in DMSO and then aliquoted into a 96-well microtiter plate. Plates were incubated at 26 °C and fungal growth was monitored by measuring an increase in absorbance (405 nm) in a BioTek Synergy4 instrument (BioTek Instruments, Inc.) for 72 h.

Statistical analysis

Differences in levels of ergosterol and plant defense metabolites were analyzed using ANOVAs, with treatment, line, and their interaction as factors contributing to observed ergosterol, phytoalexin, and other defense metabolite levels. Log transformations were deemed necessary after investigating adherence to assumptions of normality and homoscedasticity using visual inspections of quantile–quantile plots, residual diagnostics, Shapiro–Wilk’s test (normality), and Levene’s test (homoscedasticity). Post-hoc comparisons of between-treatment differences were done using Tukey’s HSD test. To investigate the importance of different metabolites in affecting observed ergosterol levels, best subsets regression, LASSO regression, and random forests (Breiman, 2001; Lumley, 2009; Tibshirani, 2011) were used to obtain variable importance measures. For best subsets regression and LASSO regression, log transformation of both metabolites and ergosterol levels was used to conform to assumptions of normality and homoscedasticity. For random forests, variable importance measures were obtained from models fit after 10 repeated 10-fold cross-validations. Rank importance measures from each method were then combined into their appropriate groups—kauralexins, zealexins, or others (SA, CA, JA, IAA, ABA, and 12-OPDA). ANOVAs with Tukey’s HSD post-hoc tests were used to evaluate differences in rank importance between classes after verifying adherence to assumptions of normality and homoscedasticity. To investigate the relationship between phytoalexins and ergosterol levels in maize kernels, exponential decay models were chosen after consideration of linear polynomial models of varying degrees based on residual diagnostic plots, log-likelihood values, and r² values. All data were collated and analyzed using R version 3.3.1 in the RStudio (version 0.99.902) development environment. Additional packages used to facilitate analysis include: xlsx for the R-Excel interface, dplyr and tidyr for data tidying, car and lsmeans for ANOVA, regression, and multiple comparisons, caret, lars, and leaps for variable importance measures, and ggplot2 for graphics.

Results

Commercial hybrid maize lines infected with three different pathogenic fungi display diverse phytoalexin defense profiles

To better understand how CHLs respond to pathogen challenge, we infected stems from nine diverse CHLs with single isolates of the common maize pathogens C. graminicola, F. graminearum, and C. heterostrophus (Fig. 1A). Seventy-two hours post-treatment, defense metabolite profiles revealed diverse concentrations of SA, CA, JA, IAA, ABA, and 12-OPDA (Fig. 1B). Comparison of grand mean averages for the nine CHLs between the C. graminicola, F. graminearum, and C. heterostrophus treatments revealed significantly lower levels of SA in F. graminearum-infected tissues than in C. graminicola- and C. heterostrophus-infected tissues (Fig. 1B; Supplementary Fig. S1A at JXB online; P<0.05). In contrast, IAA concentrations were significantly higher in F. graminearum-treated stems than in those treated with C. graminicola and C. heterostrophus (Supplementary Fig. S1D; P<0.05). Considerable differences were also observed in ABA levels, with C. heterostrophus eliciting significantly higher levels of ABA than F. graminearum and C. graminicola (Supplementary Fig. S1E; P<0.05). Interestingly, the three pathogens elicited similar levels of 13-lipoxygenase-derived oxylipins, as significant differences in JA and 12-OPDA were not observed between C. graminicola, F. graminearum, and C. heterostrophus treatments (Supplementary Fig. S1B, C; P>0.05). In quantifying fungal biomass, comparison of grand mean ergosterol averages for the nine CHLs between C. graminicola, F. graminearum, and C. heterostrophus treatments demonstrated significantly more fungal growth on C. graminicola-infected maize lines than on those infected with F. graminearum and C. heterostrophus (Fig. 1C; P<0.05). Within-treatment comparison of fungal biomass levels between individual CHLs indicated that RX813 was the most resistant line against C. graminicola infection, with an ~2- to 6-fold reduction in fungal growth compared with more susceptible CHLs (Fig. 1C). Significant differences in F. graminearum and C. heterostrophus fungal biomass were not observed between CHLs.

Fig. 1.

Commercial hybrid lines (CHLs) display diverse metabolite profiles and varied responses to C. graminicola, F. graminearum, and C. heterostrophus stem infections. (A) Representative images of slit stems 72 h post-inoculation with C. graminicola, F. graminearum, and C. heterostrophus. (B) Heat map (n=4, ± SE) displaying the maize defense metabolites salicylic acid (SA), cinnamic acid (CA), jasmonic acid (JA), auxin (IAA), abscisic acid (ABA), and 12-oxo-phytodienoic acid (12-OPDA) in nine CHLs infected with C. graminicola (Cg), F. graminearum (Fg), and C. heterostrophus (Ch). No treatment (Cont) and damage (DAM) controls are displayed as grand mean (GM) averages. (C) Quantification of fungal biomass (n=4, ± SE) measured in ergosterol across nine CHLs infected with C. graminicola, F. graminearum, and C. heterostrophus. GM averages for control, damage, and the other pathogen treatments as described above are also included. Statistical analysis was carried out within each partition; different letters a–c represent significant differences (P<0.05) by Tukey’s HSD test.

Inducibility of maize phytoalexins can be pathogen dependent

To understand the inducibility of maize terpenoid defenses in CHLs, we profiled zealexins (ZA1–ZA4 and ZB1) and kauralexins (KA1–KA3 and KB1–KB3) at 72 h after stem inoculation. Patterns of both moderately consistent and distinctly different levels of inducibility were observed among individual CHLs within specific fungal treatments, and across grand mean averages of all CHLs between fungal treatments (Fig 2A–I; Supplementary Fig. S2). As a pathogen not adapted to stem infections, C. heterostrophus was a strong general elicitor of terpenoid defenses, producing significantly higher grand mean average levels of ZA4, KA2, and KB2 than F. graminearum and C. graminicola (Fig. 2C, E, H; P<0.05). Outside of these specific examples, the common stalk rot pathogen F. graminearum elicited significantly higher total terpenoid levels (255 µg g^–1 FW) than C. heterostrophus and C. graminicola (209 µg g^–1 FW and 37 µg g^–1 FW, respectively; P<0.05). In contrast to the comparatively high levels of terpenoids accumulating in plants infected by both F. graminearum and C. heterostrophus, C. graminicola elicited significantly lower grand mean average levels of ZA1, ZB1, ZA2, ZA3, ZA4, KA1, KA3, KB1, KB2, and KB3 (Fig. 2A-J; Supplementary Fig. S2; P<0.05). Correspondingly, quantification of ergosterol levels as a measurement of fungal biomass revealed significantly more fungal growth on C. graminicola-infected plants than on F. graminearum- and C. heterostrophus-infected plants (Fig. 2J; P<0.05), consistent with a negative relationship between acidic terpenoid defenses and fungal growth.

Fig. 2.

Comparison of phytoalexin production and fungal growth across nine commercial hybrid lines (CHLs) infected with C. graminicola, F. graminearum, and C. heterostrophus. Average (n=4, ± SE) fungal-induced zealexin A1 (A), B1 (B), A4 (C), and kauralexin, A1 (D), A2 (E), A3 (F), B1(G), B2(H), and B3 (I) concentrations 72 h post-stem infection with C. graminicola, F. graminearum, and C. heterostrophus across individual CHLs. Grand mean (GM) averages for control, damage, and the other pathogen treatments as described above are also included. Statistical analysis was carried out within each partition; different letters a–e represent significant differences (P<0.05) by Tukey’s HSD test. (J) Grand mean averages (n=4, ± SE) of ergosterol and total phytoalexins in stems of all nine CHLs under no-treatment control, damage, C. graminicola, F. graminearum, and C. heterostrophus treatments. Points and error bars denote means and bootstrapped 95% confidence intervals, respectively. Groups within the same color not sharing the same letter are significantly different (Tukey HSD test; P<0.05). HSD test.

Select terpenoids with differential accumulation patterns display diverse antimicrobial activity in fungal growth assays

The phytoalexins ZA4, KA2, and KB2 were only modestly elicited by C. graminicola and F. graminearum infection. To assess the antimicrobial activity of these three metabolites, we compared in vitro growth curves of C. graminicola, F. graminearum, and C. heterostrophus using nutrient media amended with either ZA4 or a mixture of KA2/KB2 at either 0, 10, 50, or 100 µg ml^–1, representing concentrations within a reasonable physiological range (Huffaker et al., 2011; Schmelz et al., 2011; Christensen et al., 2017). ZA4 and KA2/KB2 had significant inhibitory activity against F. graminearum, as both sequentially reduced fungal growth at each stepwise increase in dose spanning 10, 50, and 100 µg ml^–1 concentrations (Fig. 3A, P<0.05; Fig. 3B, D, P<0.05). At 50 µg ml^–1 and 100 µg ml^–1, KA2/B2 activity also resulted in a significant reduction in C. graminicola growth; however, 10 µg ml^–1 promoted greater fungal growth than controls (Fig. 3E; P<0.05). While KA2/B2 also had significant yet modest inhibitory effects against C. heterostrophus (Fig. 3C; P<0.05), ZA4 stimulated C. heterostrophus growth at all doses examined (Fig 3D). Curiously, C. graminicola growth was also stimulated to significantly higher levels in liquid culture with ZA4 than DMSO controls at all concentrations, an effect that displayed little dose dependency (Fig. 3F; P<0.05).

Fig. 3.

Fungal growth in response to zealexin (ZA4) and kauralexin A2/B2 (KA2/KB2) activity. Average (n=8, ± SE) F. graminearum (A and B), C. heterostrophus (C and D), and C. graminicola (E and F) growth (OD_{450 nm}) in liquid culture containing nutrient broth with KA2/KB2 and ZA4 at concentrations of 0 (white), 10 (light gray), 50 (dark gray), and 100 (black) µg ml^–1. Within plots, different letters a–d represent significant differences (P<0.05 for all ANOVAs; Tukey test corrections for multiple comparisons: P<0.05).

Rank of importance and genetic evidence support the role of microbe- and drought-inducible terpenoids in defense against plant pathogens

Bioassay results (Fig. 3A–F) support the hypothesis that some defense metabolites can have greater defensive properties than others. To investigate the association of individual metabolites with fungal biomass estimates measured via ergosterol levels, we utilized three statistical models—best subsets regression, LASSO regression, and random forests—to determine the relative importance of each metabolite on influencing ergosterol levels (Breiman, 2001; Lumley, 2009; Tibshirani, 2011). Of the defense-related metabolites analyzed across nine CHLs infected with three fungal pathogens (C. graminicola, F. graminearum, and C. heterostrophus), KA3 was ranked as most important by each algorithm (Fig. 4A). Three other kauralexins were ranked in the top four, namely KA1, KA2, and KB1. Of the zealexins, ZA4 was ranked third and ZA1 was ranked fourth in importance by random forests modeling. Comparison of total kauralexins, zealexins, and other measured metabolites (SA, CA, JA, IAA, ABA, and 12-OPDA) revealed a significantly higher ranking for kauralexins than the other two groups (Fig. 4B; P<0.05). Collectively, these results are supportive of an endogenous defensive role for inducible terpenoids in protecting against common maize pathogens.

Fig. 4.

Maize kauralexins play an important role in pathogen defense. (A) Top four results from the comparison of individual metabolites for ranks of importance in affecting fungal growth in maize stems. Ranks were identified by best subsets regression, LASSO regression, and random forests models across the combined data of all nine CHLs infected with C. graminicola, F. graminearum, and C. heterostrophus for 72 h. Compounds are presented in their respective classes: kauralexins (KA), zealexins (ZA), or other defense metabolites [salicylic acid (SA), cinnamic acid (CA), jasmonic acid (JA), auxin (IAA), abscisic acid (ABA), and 12-oxo-phytodienoic acid (12-OPDA)]. (B) Comparison of means for ranks of importance between kauralexins, zealexins, and other defense metabolites in affecting fungal growth. Lower values indicate greater relative importance. Points and error bars denote the mean and SE, respectively. Different letters indicate significant differences (P<0.05; Tukey’s HSD test). (C and D) Analysis of wild-type W22 and an2 mutants under control (cont), damage (Dam), C. graminicola (Cg), F. graminearum (Fg), and C. heterostrophus (Ch) treatments. Quantifications (n=4, ± SE) of total kauralexins (C) and fungal growth measured in ergosterol (D) were carried out 72 h post-inoculation (ANOVA: asterisk represents significant differences between genotypes; **P<0.01; ***P<0.001).

With kauralexins showing the highest rank of importance, we genetically investigated their roles on C. graminicola, F. graminearum, and C. heterostrophus growth by utilizing the maize mutant an2 lacking kauralexin biosynthetic capacity (Vaughan et al., 2015). Measurement of kauralexins in fungal-infected tissues of the corresponding W22 WT showed induction patterns similar to those observed in the nine CHLs, with kauralexin levels notably higher in C. heterostrophus- and F. graminearum-infected tissues compared with those infected with C. graminicola (Fig. 4C). As expected, an2 mutants were functionally devoid of kauralexins (Fig. 4C). Analysis of ergosterol levels in pathogen-infected WT and an2 mutant stems demonstrated that an2 mutants were significantly more susceptible to C. heterostrophus (Fig. 4D). While ergosterol levels trended higher in an2 mutant stems infected with F. graminearum, there were no significant differences at the time point analyzed.

Interestingly, the drought- and pathogen-inducible phytohormone ABA obtained the second highest ranking in two of three algorithms testing association with pathogen resistance (Fig. 4A). To investigate the role of ABA further, we inoculated the stems of B73 WT and ABA-deficient vp14 mutant plants with the pathogen consistently promoting the greatest accumulation of ABA (i.e. C. heterostrophus). Although vp14 mutants were significantly reduced in ABA production (P<0.01; Supplementary Fig. S3), significant changes in ergosterol levels associated with C. heterostrophus infection were not observed

Fusarium verticillioides kernel infection reveals diverse levels of CHL resistance and a negative relationship between acidic terpenoid accumulation and fungal growth

A naturally rich source of terpenoid phytoalexins are elicited scutella tissues present in maize kernels (Huffaker et al., 2011; Schmelz et al., 2011). Given this biosynthetic potential, we investigated kauralexin and zealexin kernel production in the nine CHLs in response to the ear-rot pathogen F. verticillioides. At 5 days post-inoculation, we observed a wide range of resistance and susceptibility across the CHLs with 3- to 9-fold differences in ergosterol existing between the most resistant (DK415 and DK580) and most susceptible (DK657 and RX813) lines (Fig. 5A). Comparison of total zealexin and kauralexin levels with ergosterol suggested a negative relationship between phytoalexin production and fungal biomass (Fig. 5B; Supplementary Fig. S4). Examination of these data by regression analyses supported a negative correlation between phytoalexin production and fungal growth (P<0.0001; R²=0.30). In vitro bioassays of individual phytoalexins in liquid culture demonstrated that ZA1, KB2, and KA2 all had significant growth-inhibitory activity against F. verticillioides at the 100 µg ml^–1 concentration (Fig. 5C).

Fig. 5.

Phytoalexin concentrations negatively correlate with fungal growth. (A) Kernel assay phenotypes (n=3) from maize CHLs 5 d post-inoculation with the ear-rot pathogen Fusarium verticillioides. Bars and error bars denote mean ergosterol levels and 95% bootstrapped confidence intervals, respectively. (B) Regression analysis investigating the relationship between total phytoalexin concentrations and F. verticillioides growth in infected CHL kernels. The line and shaded gray area indicate exponential decay model fit and 95% confidence intervals, respectively (P<0.0001; R²=0.3). (C) Average (n=8, ± SE) F. verticillioides growth based on absorbance (OD_{405 nm}) following 72 h in nutrient broth at 100 µg ml^–1 concentrations of ZA1, KA2, and KB2 dissolved in DMSO. Within plots, different letters a–c represent significant differences (P<0.05 for all ANOVAs; Tukey test corrections for multiple comparisons: P<0.05).

Genetic evidence demonstrates that kauralexins protect maize kernels against F. verticillioides infection

Given the negative correlation between terpenoid phytoalexin production and F. verticillioides growth in the nine CHLs, we hypothesized that maize kernels deficient in kauralexins would display increased F. verticillioides growth. Five days after kernel inoculation, an2 mutants displayed greater susceptibility to F. verticillioides infection than WT W22 kernels based on a simple visual inspection (Fig. 6A). As expected, kauralexins were absent in mock-inoculated and F. verticillioides-infected an2 mutants, but were strongly induced in infected W22 samples (Fig. 6B). As a quantitative estimate of fungal biomass, ergosterol levels were consistent with a 3-fold increase in observed fungal growth on an2 mutants compared with W22 (P<0.001; Fig. 6C). Furthermore, under field conditions conducive to naturally occurring F. verticillioides ear-rot, a very similar pattern was observed, where an2 mutants displayed heightened disease symptoms compared with WT W22 ears (Supplementary Fig. S5). Collectively, the results of our pharmacological assays, biochemical analyses, endogenous relationships, and mutant analyses under laboratory and natural field conditions demonstrate that kauralexins play an important role in the protection of maize tissues against fungal attack.

Fig. 6.

Genetic investigation demonstrates a significant defensive role for kauralexins against F. verticillioides. (A) Kernel assay phenotypes from wild-type W22 and an2 mutants 5 d post-inoculation with the ear-rot pathogen F. verticillioides. (B and C) Quantification (n=4 ± SE) of total kauralexins and ergosterol in W22 and an2 mutant kernels from damage (DAM) and F. verticillioides (F.vert) infection treatments (ANOVA: asterisk represents significant differences between genotypes; **P<0.01; ***P<0.001).

Discussion

Studies examining crop defense mechanisms that protect against fungal pathogens are generally viewed as broadly applicable. However, projections are typically based upon a narrow selection of research lines that could be substantially different from those in commercial production. Moreover, certain antifungal terpenoid defenses are commonly missing altogether in diverse inbred maize lines (Ding et al., 2017). With so little public knowledge available regarding the specific genetics underlying CHLs, it is difficult to predict which traits (i.e. genes/biochemical pathways) have been intentionally selected for and included. To investigate the defensive architecture in available commercial lines, we analyzed a panel of nine diverse CHLs infected with four relevant fungal pathogens, namely C. graminicola, F. graminearum, C. heterostrophus, and F. verticillioides. Interestingly, we observed relatively low variation in elicited defense metabolite production within the stem tissues of different CHLs infected with the same pathogen. In contrast, distinct and consistent differences were observed in CHL defense responses to different fungal pathogens. In one such example, F. graminearum elicited low levels of SA compared with C. graminicola and C. heterostrophus. This could be related to the contrastingly high levels of JA produced in F. graminearum-infected tissues, as jasmonates have been reported to antagonize the SA pathway (Glazebrook, 2005; Thaler et al., 2012). Indeed, JA levels were highest in F. graminearum-infected tissues, with an average 32% higher concentration than in C. graminicola- and C. heterostrophus-infected tissues, further supporting a role for JA signaling in terpenoid biosynthesis (Huffaker et al., 2011; Schmelz et al., 2011). Major differences were also observed with the drought-inducible hormone ABA, being dramatically reduced in C. graminicola-infected tissues compared with those infected with F. graminearum and C. heterostrophus.

In the analysis of direct defense metabolites, F. graminearum infection also elicited high levels of established zealexins, namely ZA1 and ZB1, but accumulation of ZA4 was dramatically reduced to levels more comparable with those elicited by C. graminicola. This was curious considering that ZA4 was strongly elicited by C. heterostrophus (Fig. 2C), Aspergillus flavus, and Rhizopus microsporus (Christensen et al., 2017). In liquid culture bioassays, ZA4 demonstrated strong antimicrobial activity against F. graminearum (Fig. 3B). Collectively, our results indicate that F. graminearum infection and pathogenicity may benefit from the reduced elicitation of ZA4 in planta. Moreover, the results suggest that a F. graminearum effector might exist, capable of suppressing specific maize terpenoid biosynthetic pathways that would otherwise enhance plant defense. Similar observations were previously seen in a study where the F. graminearum effector FGL1 was found to suppress particular defensive phytoalexins in pea and bean (Blumke et al., 2014).

Unlike evidence for highly selective defense suppression following F. graminearum infection, C. graminicola demonstrated a widespread suppression of defenses with significantly lower levels in almost every terpenoid measured. Defense suppression was inversely related to total ergosterol levels, demonstrating that C. graminicola-infected tissues produced significantly more fungal biomass than those infected with F. graminearum and C. heterostrophus (Fig. 2J). Moreover, strong antimicrobial activity of kauralexin diacids against C. graminicola in liquid culture assays (Fig. 3) is consistent with pathogenicity benefits for C. graminicola in an environment with suppressed inducible defenses (Schmelz et al., 2011). Colletotrichum graminicola effector-driven benefits have been previously observed in maize where a mutation in the fungalysin metalloprotease effector increased maize resistance (Sanz-Martin et al., 2016). While changes in chitinase activity were associated with modified resistance, acidic terpenoid production may also play a role, as pathogen infection strongly elicits complex processes that increase metabolite and defense protein levels (Huffaker et al., 2011).

In contrast to C. graminicola, C. heterostrophus was a strong elicitor of stem defenses, inducing appreciable levels of all acidic terpenoids with significantly higher ZA4 and kauralexin diacid production than C. graminicola and F. graminearum. In liquid culture bioassays, the KA2/KB2 mixture displayed significant antimicrobial activity against C. heterostrophus; however, ZA4 had a stimulatory effect on C. heterostrophus growth, an effect also observed with C. graminicola. As highly inducible terpenoids are generally anticipated to have broad antimicrobial activities, the growth-promoting effect of ZA4 on C. heterostrophus and C. graminicola was unexpected. However, this finding is consistent with earlier results demonstrating a stimulatory effect of phenylpropanoid-derived defenses on fungal spore germination (Ruan et al., 1995). It is possible that C. heterostrophus may have the ability to detoxify, tolerate, or even utilize several terpenoids including ZA4. Given both growth-promoting and suppressive activities, complex combinatorial outcomes suggest that widespread defense suppression is not required for significant C. heterostrophus pathogenicity. Despite the complexities observed with select zealexins in vitro, the large-scale removal of kauralexins in an2 mutants facilitated increased C. heterostrophus growth. This demonstrates that kauralexins, in the context of diverse defenses present in vivo, significantly promote fungal growth suppression in planta.

To determine the comparative impact of kauralexins with other classes of defense metabolites in affecting observed ergosterol levels, we analyzed the association of kauralexins, zealexins, and other defense metabolites (SA, CA, JA, IAA, ABA, and 12-OPDA) with reduced fungal growth. Statistical models in maize stems highlight kauralexins as contributing more greatly to fungal growth suppression than any other class of defense compounds (Fig. 4A). Comparison of kauralexins and zealexins in this and previous studies affirm these findings. For example, ZA4 had strong antimicrobial activity against F. graminearum, but the KA2/KB2 mix demonstrated a broader negative impact across all three fungi (Fig. 3). Antimicrobial assays with R. microsporus showed a 75% reduction in fungal growth in the presence of KA3 (Schmelz et al., 2011), as compared with a 45% reduction with ZA1 (Huffaker et al., 2011). Moreover, these collective results strongly align with statistical ranking of individual metabolites, as KA3 was determined to contribute the most to fungal growth suppression, whereas ZA1 was less impactful (Fig. 4A). While kauralexins have great potential in defense against C. graminicola, the consistently strong suppression of kauralexin production by this pathogen largely prevents detection of increased fungal growth in an2 mutants. Statistically insignificant fungal growth differences between WT and an2 mutants in F. graminearum-infected tissues were more unexpected. While several kauralexins are strongly inducible upon F. graminearum infection, the pathogen undoubtedly contains detoxification mechanisms to manage the endogenous production and catabolism of terpenoid-based mycotoxins (Kimura et al., 1998; Ohsato et al., 2007). Collectively, these results suggest that kauralexins play a predominant role in pathogen defense but that zealexins may play a more prominent role against F. graminearum.

Curiously of all the metabolites quantified, statistical analyses ranked ABA with the second highest level of importance for affecting fungal biomass as estimated by ergosterol levels. The phytohormone ABA is commonly known for its role in response to drought (Jia et al., 2001) and defense against pathogens (Adie et al., 2007; Ton et al., 2009), including recent evidence demonstrating its capacity to elicit acidic terpenoid accumulation in maize (Vaughan et al., 2015). ABA levels are also more strongly induced by C. heterostrophus and F. graminearum compared with C. graminicola, a general pattern that mirrors terpenoid phytoalexin production in response to these pathogens. Despite indicators suggesting a significant role for ABA in defense against fungi, only marginal differences in fungal growth were observed between the WT and ABA-deficient vp14 mutants (Supplementary Fig. S3B). Substantial differences in concentrations of ABA between C. heterostrophus-infected WT plants in the B73 background (~100 ng g^–1 FW) and those produced in commercial hybrid backgrounds (~1.4 µg g^–1 FW) may explain the variation between the statistical evaluation of the CHLs and our genetic analysis. Similar to results previously reported in other plant species (Asselbergh et al., 2008; Flors et al., 2009), these data imply that ABA-mediated resistance may be concentration dependent, a hypothesis that will require further investigation in maize.

To address interactions in additional tissues, we screened the nine CHLs for terpenoid defense responses in kernels following F. verticillioides inoculation. Kernels are a predictable site of meaningful interactions, as previous studies demonstrated the abundance of acidic terpenoids in scutella tissues (Huffaker et al., 2011; Schmelz et al., 2011; Christensen et al., 2017). Diverse disease symptoms across CHLs showed sizeable differences in fungal growth and a general pattern for low ergosterol levels in lines showing heightened acidic terpenoid accumulation. The results derived from the exponential decay model (Fig. 5B) estimate that ~32% of the variability in F. verticillioides growth is explained by phytoalexin concentrations, consistent with a predicted biological role in reducing levels of F. verticillioides infection. The antimicrobial activity of zealexins and kauralexins against F. verticillioides was further confirmed in liquid cultures with significant reduction in F. verticillioides growth under ZA1, KA1, and KA2 treatments. To substantiate the functional role of kauralexins in defense against F. verticillioides, we infected WT W22 and an2 mutant kernels and observed clear differences in fungal growth between the two genotypes, including a 3-fold increase in ergosterol in an2 mutants. A similar result was observed under natural field conditions, as an2 mutants appeared highly susceptible to F. verticillioides ear-rot compared with WT W22. Collectively, these results demonstrate an important biological function for kauralexins in protecting maize kernels from ear-rotting pathogens such as F. verticillioides.

In response to pathogen attack, maize contains a complex blend of small molecule defenses originating from diverse biosynthetic pathways (e.g. benzoxazinoids, phenylpropanoids, oxylipins, and terpenoids) (Huffaker et al., 2011; Balmer et al., 2013; Schmelz et al., 2014; Christensen et al., 2015; Ding et al., 2017). While typically analyzed by targeted methods, the actual biochemical arrays are exceedingly complex mixtures that combine to function in plant defense. In our in vitro bioassays examining the action of acidic terpenoids on diverse pathogenic fungi, we uncovered both positive and negative growth effects that were dependent on the fungal species. The complexity of known plant defenses coupled with variable microbial responses to single chemicals suggests that a meaningful understanding of maize resistance can be better demonstrated in the context of defined mutations that block individual enzymatic steps. Bypassing these challenges, our current examination of an2 mutants demonstrates that genetic deletion of inducible kauralexin production promotes measurable increases in susceptibility to C. heterostrophus and F. verticillioides. These results provide an empirical mechanistic demonstration in planta and answer the long-standing question regarding whether or not fungal-induced transcripts of An2 mediate meaningful maize defenses (Harris et al., 2005). In the near future, additional approaches will be required to define many of the genetic nodes controlling biosynthetic and regulatory pathways contributing to maize defense. An understanding of the underlying genes encoding enzymes in biochemical defense pathways will provide an additional toolbox to enhance targeted breeding approaches and bolster maize resilience to biotic threats (Kollner et al., 2008; Degenhardt et al., 2009; Ding et al., 2017).

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Defense-related metabolite production in response to C. graminicola, F. graminearum, and C. heterostrophus stem infections.

Fig. S2. Comparison of zealexin A2 and A3 production across nine commercial hybrid lines (CHLs) infected with C. graminicola, F. graminearum, and C. heterostrophus.

Fig. S3. Genetic investigation of ABA resistance against C. heterostrophus stem infections.

Fig. S4. Quantification (n=4, ± SE) of total zealexins and kauralexins in the nine commercial hybrid lines 5 d post-inoculation of kernels with F. verticillioides.

Fig. S5. an2 mutants are more susceptible to F. verticillioides under natural field conditions.

Acknowledgements

We would like to thank the Monsanto company for graciously allowing us to use their commercial hybrid lines for this study. We thank Dawn Diaz-Ruiz, Erika Friman, Amanda Balon, Bruce Schnicker, Steve Willms, Bevin Forguson, and Maritza Romero for their technical support. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA. Research was funded by the US Department of Agriculture (USDA) Agricultural Research Service Projects 6615-21000-010-00/6615-22000-027-00, and by an NSF Division of Integrative Organismal Systems Competitive Award 1139329.