The Genetics of Leaf Flecking in Maize and Its Relationship to Plant Defense and Disease Resistance1[OPEN] Bode A. Olukolu, Yang Bian, Brian De Vries, William F. Tracy, Randall J. Wisser, James B. Holland, and Peter J. Balint-Kurti* Department of Plant Pathology, North Carolina State University, Raleigh, North Carolina 27695–7616 (B.A.O., P.J.B.-K.); Department of Crop Science, North Carolina State University, Raleigh, North Carolina 27695–7620 (B.A.O., Y.B., J.B.H.); Department of Agronomy, University of Wisconsin, Madison, Wisconsin 53706 (B.D.V., W.F.T.); Department of Plant and Soil Sciences, University of Delaware, Newark, Delaware 19716 (R.J.W.); and United States Department of Agriculture-Agricultural Research Service Plant Science Research Unit, Raleigh, North Carolina 27695 (J.B.H., P.J.B.-K.) ORCID IDs: 0000-0003-4143-8909 (B.A.O.); 0000-0003-1075-0115 (R.J.W.); 0000-0002-4341-9675 (J.B.H.); 0000-0002-3916-194X (P.J.B.-K.).

Physiological leaf spotting, or flecking, is a mild-lesion phenotype observed on the leaves of several commonly used maize (Zea mays) inbred lines and has been anecdotally linked to enhanced broad-spectrum disease resistance. Flecking was assessed in the maize nested association mapping (NAM) population, comprising 4,998 recombinant inbred lines from 25 biparental families, and in an association population, comprising 279 diverse maize inbreds. Joint family linkage analysis was conducted with 7,386 markers in the NAM population. Genome-wide association tests were performed with 26.5 million single-nucleotide polymorphisms (SNPs) in the NAM population and with 246,497 SNPs in the association population, resulting in the identification of 18 and three loci associated with variation in flecking, respectively. Many of the candidate genes colocalizing with associated SNPs are similar to genes that function in plant defense response via cell wall modification, salicylic acid- and jasmonic acid-dependent pathways, redox homeostasis, stress response, and vesicle trafficking/remodeling. Significant positive correlations were found between increased flecking, stronger defense response, increased disease resistance, and increased pest resistance. A nonlinear relationship with total kernel weight also was observed whereby lines with relatively high levels of flecking had, on average, lower total kernel weight. We present evidence suggesting that mild flecking could be used as a selection criterion for breeding programs trying to incorporate broad-spectrum disease resistance.

The plant hypersensitive response (HR) is a form of programmed cell death (PCD) characterized by rapid, localized cell death at the point of attempted pathogen penetration, usually resulting in disease resistance (Coll et al., 2011). It is often associated with other responses, including ion fluxes, an oxidative burst, lipid peroxidation, and cell wall fortification (Hammond-Kosack and Jones, 1996). van Doorn et al. (2011) suggested that HR is a type of PCD sharing features with, but distinct from, both vacuolar cell death and necrosis.

1 This work was supported by the U.S. Department of AgricultureAgricultural Research Service and by the National Science Foundation (grant nos. 0822495, 1127076, 1238014, and 1444503). * Address correspondence to [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Peter J. Balint-Kurti ([email protected]). P.J.B.-K. and B.A.O. conceived the project; B.A.O., W.F.T., B.D.V., and P.J.B.-K. planned the field experiments and took the field data; B.A.O. carried out all the other experiments; B.A.O., Y.B., R.J.W., and J.B.H. performed the data analysis; B.A.O., J.B.H., and P.J.B.-K. wrote the article. [OPEN] Articles can be viewed without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.15.01870

HR has been associated with resistance to almost every class of pathogen and pest, including bacteria, viruses, fungi, nematodes, insects, and parasitic plants (Wu and Baldwin, 2010), and generally is most effective against biotrophic pathogens, since biotrophs require a long-term feeding relationship with living host cells. It is generally mediated by dominant resistance (R) genes whose activation is triggered by the direct or indirect detection of specific pathogen-derived effector proteins (Bent and Mackey, 2007). R proteins are maintained in their inactive state if their corresponding effector is not present. Mutants in which HR is constitutively active have been identified in many plant species, including maize/corn (Zea mays; Walbot et al., 1983; Johal, 2007), Arabidopsis (Arabidopsis thaliana; Lorrain et al., 2003), barley (Hordeum vulgare; Wolter et al., 1993), and rice (Oryza sativa; Yin et al., 2000). One well-known class of plant mutants spontaneously form lesions (patches of dead or chlorotic cells) in the absence of any obvious injury, stress, or infection to the plant. Since these lesions in some cases resemble HR, they have been termed disease-lesion mimics (Neuffer and Calvert, 1975). These mutants, which we will here collectively term Les mutants, have been studied extensively, especially in maize (Walbot et al., 1983; Johal et al., 1995; Johal, 2007) and Arabidopsis

Plant PhysiologyÒ, November 2016, Vol. 172, pp. 1787–1803, www.plantphysiol.org Ó 2016 American Society of Plant Biologists. All Rights Reserved. 1787

Olukolu et al.

(Coll et al., 2011). While some of these lesion phenotypes are indeed caused by perturbations in the plant defense response (Hu et al., 1996; Rustérucci et al., 2001), some of the genes underlying this mutant class affect various other pathways that cause cell death if their function is perturbed (Johal, 2007). For instance, the Arabidopsis gene acd2 and the maize gene lls1 are defective in chlorophyll degradation (Gray et al., 1997; Mach et al., 2001). We have defined leaf flecking as the mild, genetically determined spotting observed on many maize inbred cultivars (Vontimitta et al., 2015; Fig. 1). The trait is qualitatively and visually similar to, but quantitatively less severe than, Les mutant phenotypes. The distinction between what constitutes a flecking versus a mild Les trait is necessarily somewhat arbitrary, but for our purposes, we have defined any nonproliferating and distinct leaf-spotting phenotype as flecking. Leaf flecking is familiar to most corn breeders, appearing in such well-known and widely used lines such as Mo17 (Zehr et al., 1994) and in several other species such as barley (Makepeace et al., 2007), wheat (Triticum aestivum; Nair and Tomar, 2001), and oat (Avena sativa; Ferdinandsen and Winge, 1930). Flecking tends to be more noticeable in inbreds compared with their derived hybrids (M. Goodman and W. Dolezal, personal communication). Anecdotally, it is often thought to be indicative of a constitutive low-level defense response and as a marker for increased disease resistance. In previous work, we and others have defined the genetic architectures associated with resistance to several maize diseases, including southern leaf blight (SLB; causal agent, Cochliobolus heterostrophus), northern leaf blight (NLB; causal agent, Exserohilum turcicum), and gray leaf spot (GLS; causal agent, Cercospora zeae-maydis;

Kump et al., 2011; Poland et al., 2011; Wisser et al., 2011; Benson et al., 2015), and with the control of the maize HR (Chintamanani et al., 2010; Chaikam et al., 2011; Olukolu et al., 2013). For much of this work, we used two powerful mapping populations: the maize association population (Flint-Garcia et al., 2005), a collection of 302 diverse inbred lines with low linkage disequilibrium, and the 5,000-line nested association mapping (NAM) population (McMullen et al., 2009), which is made up of 25 200-line recombinant inbred line (RIL) subpopulations derived from crosses between the common parent B73 and 25 diverse inbreds. Using these populations, it is possible to both sample a diverse array of germplasm and map quantitative trait loci (QTLs) precisely, in some cases to the gene level (Tian et al., 2011; Cook et al., 2012; Hung et al., 2012; Larsson et al., 2013; Olukolu et al., 2013; Wang and Balint-Kurti, 2016). A recent study using 300 lines from the maize intermated B73 3 Mo17 population advanced intercross line mapping population identified low but moderately significant positive correlations between increased flecking and increased disease resistance and defense response (Vontimitta et al., 2015). Loci associated with variation in flecking were mapped, although these loci did not colocalize with QTLs identified previously for disease resistance and defense response traits (BalintKurti et al., 2007, 2008, 2010; Olukolu et al., 2013). In this study, we have extended this work to examine the genetic basis of leaf flecking over a much more diverse set of maize germplasm using a substantially larger population. We mapped loci associated with variation in leaf flecking and identified candidate genes and pathways that may be involved in this phenotype. Additionally, we have examined the correlations between leaf flecking and disease resistance, the hypersensitive defense response, and total kernel weight.

Figure 1. A, Examples of variation in the flecking phenotype among inbred lines, with severity increasing from left to right (flecking scores in parentheses, from 0 to 4, scored on a scale of 1–10). B, Leaves of the lines nearly isogenic to inbred Mo20W, into which specific indicated dominant Les mutant genes have been introgressed (Rp1-D21 mutation in an H95 inbred background). Photographs were taken in Clayton, North Carolina, 12 weeks after planting. This figure is adapted from Figure 1 of Vontimitta et al. (2015).

1788

Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

RESULTS Evaluation of the Maize Flecking Phenotype

The NAM and association populations were evaluated in two and five environments, respectively, and scored on a scale of 0 to 10, with 0 indicating no flecking and 10 indicating 100% necrosis (Fig. 1). Since the flecking phenotype is relatively mild, we did not score any lines higher than 6 on this scale. Heritability in the NAM population was estimated at 0.86 and 0.93 on plot and line mean bases, respectively, while estimates in the association population were 0.51 and 0.86 on plot and line mean bases, respectively. The average flecking standardized area under the curve (sAUC) scores were 1.1 (maximum, 5.7) and 1.3 (maximum, 7.2) for the NAM and association populations, respectively (Supplemental Figs. S1 and S2; Supplemental Files S1 and S2). For the association population, correlations were highly significant for every pairwise combination of environments (Supplemental Table S1). The two controlled environments (greenhouse and growth chamber) were the most highly correlated (0.85), while for the field environments, the two Clayton, North Carolina, environments (NC12 and NC13) were the most highly correlated (0.77; Supplemental Table S1). The correlation coefficient between NAM line scores across the two environments was 0.51 (P , 0.0001). Flecking QTLs in the NAM Population

Joint-family (JF) and single-family (SF) linkage analyses were performed using the best linear unbiased predictors (BLUPs) values for flecking derived from the NAM population. The JF linkage analysis identified 17 QTLs (Figs. 2 and 3; Supplemental Fig. S3). The lengths of the JF QTL support intervals ranged from 1.8 cM (0.71 Mb) to 10 cM (92.04 Mb), with a mean of 4.81 cM (15.85 Mb; Supplemental Files S3–S6). SF analyses identified two to seven QTLs segregating in each family (Figs. 2 and 3), associated with 4.1% to 17.7% of the variance among lines within the family. The total phenotypic variance within a family associated with all mapped QTLs ranged from 14% to 54.5% among families. Most of the JF QTLs segregated in more than one SF (Fig. 3). A comprehensive search for digenic epistatic QTL interactions based on the joint and single linkage model did not identify any significant epistatic interactions. NAM and Association Population Genome-Wide Association Analysis, Candidate Genes, and Associated Pathways

Association analyses were performed in both NAM and association populations. To determine significant single-nucleotide polymorphisms (SNPs) in NAM, a resampling analysis was conducted in which 80% of the lines within each family were sampled and subjected to Plant Physiol. Vol. 172, 2016

genome-wide association (GWA) analysis, and this process was repeated 100 times. The proportion of data samples in which an SNP is detected as significantly associated with the trait is the resample model inclusion probability (RMIP) for that SNP. An RMIP threshold of 0.25 was used to declare significant associations in NAM . For the association population, a false discovery rate test-based threshold of 2.1 3 1025 (2log10 P = 5.1) and a permutation test (2log 10 P = 5.46) were used to establish significance. A diagnostic Q-Q plot (Supplemental Fig. S4) reflects that the marker-trait association was well controlled for population structure (i.e. low false-positive rate). Significant associations were detected at 18 loci in the NAM (Supplemental Fig. S5) and three loci in the association (Supplemental Fig. S6) panel, respectively (Supplemental Files S7–S9). In the NAM population, eight of the 18 associated loci colocalized with the JF flecking QTLs, while 14 associated loci colocalized with the SF flecking QTLs (Supplemental File S10). For seven of the eight colocalizing SNP and QTL pairs, the SNP effect was in the same direction as the average QTL effect (Supplemental Fig. S7). Two of the four loci identified from the association population colocalized with two SF flecking QTLs (Supplemental File 11). The closest annotated genes to the associated SNPs, hereafter referred to as candidate genes, were identified (Table I). In the NAM population analysis, of the 18 associated SNPs, six, three, and nine associated loci were localized within the coding sequence, less than 5 kb from the coding sequence, and between 5 and 100 kb from the coding sequence of the nearest gene, respectively (Supplemental File S10). In the association population analysis, one associated SNP was localized within the coding sequence, while the two other associated loci were localized 637 and 3,248 bp downstream of the candidate genes (Supplemental File S11). Using literature-based annotations, the candidate genes identified were predicted to play roles in processes associated with defense response, redox homeostasis, stress response, cell wall modification, and vesicle trafficking/remodeling (Table I). Effect of Light on Flecking Severity

Light modulates the severity of various Les mutants (Gray et al., 2002; Penning et al., 2004; Negeri et al., 2013). To establish if variation in light intensity had an effect on the severity of flecking, moderately, mild, and nonflecking lines (Mo17, Ki11, and NC250, respectively) were grown under low light intensity (0.09–0.12 mE m22 s21) in the greenhouse until 4 weeks old (lowlight treatment). At 4 weeks, half the plants from each line were transferred to a high-light-intensity (0.28–0.3 mE m22 s21) environment keeping all other conditions constant. The plants were grown for another week. Flecking was not observed on NC250 leaves under either light condition. Ki11 displayed no flecking and mild flecking under low- and high-light regimes, 1789

Olukolu et al.

Figure 2. Heat map showing the additive allelic effects flecking trait across 25 NAM founder lines relative to the common B73 parent. Asterisks indicate that the effect was statistically significant. Chromosome and genetic map positions (centimorgan [cM]) of QTL peaks are shown on the left vertical column, and the NAM founder lines are shown on the top horizontal row. Map positions are based on the NAM genetic map (McMullen et al., 2009). The scale bar below the heat map indicates the range of allelic effect values and corresponding color intensity.

respectively. Mo17 had very mild flecking and moderate flecking in low- and high-light regimes, respectively (Fig. 4).

and Oh7b; Fig. 5). ROS levels in the moderate flecking line NC358 were similar to those in the nonflecking lines.

Correlation of Reactive Oxygen Species Levels with the Flecking Phenotype

Correlation of Flecking with HR-Related, Disease Resistance, and Yield Traits

We measured total reactive oxygen species (ROS) levels in 2-, 3-, and 4-week-old inbred lines, which displayed variable flecking phenotypes and included nonflecking (score of 0), mild flecking (chlorotic spots; score of 0.5–3), and moderate flecking (necrotic spots; score of 3–5) lines. Low ROS levels were consistently detected in the nonflecking lines, with significantly elevated levels detected in the lines with low flecking. ROS levels in three of the four lines with moderate flecking were intermediate between the mild flecking lines with higher ROS levels (Ki11 and Tx303) and the nonflecking lines with lower ROS levels (B97, CML69,

Rank correlation coefficients on both population structure adjusted and unadjusted least squares mean (LSmean) values were estimated between flecking and several traits that had been measured previously on one or both populations: HR severity induced by the autoactive resistance gene Rp1-D21 (Olukolu et al., 2013, 2014), resistance to SLB (Kump et al., 2011; Wisser et al., 2011), resistance to NLB (Poland et al., 2011; Wisser et al., 2011), resistance to GLS (Wisser et al., 2011; Benson et al., 2015), resistance to Fusarium ear rot (Fusarium verticillioides; Zila et al., 2013), resistance to Mediterranean corn borer (MCB; Samayoa et al., 2015),

1790

Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

Figure 3. QTLs obtained from SF and JF QTL analyses for flecking across all 10 maize chromosomes/linkage groups. The genetic distance for each chromosome is represented in cM on the horizontal axis. QTL confidence intervals are shown as thick horizontal bars. Each SF analysis is designated by the non-B73 parental inbred of each family shown on the vertical axis. QTLs identified from JF linkage mapping in the NAM population comprising all 25 populations are indicated on the bottom row.

common rust resistance (Olukolu et al., 2016), and total kernel weight (Hung et al., 2012). All the traits had been scored over multiple environments across both the NAM population and the association population except for resistance to Fusarium ear rot, common rust, and MCB, which were scored only in the association population. In both adjusted and unadjusted values, higher severity of flecking was mostly consistent and significantly correlated with more severe HR and higher levels of resistance to the diseases NLB, GLS, and SLB in both the NAM and association populations, except for the negative correlation (population structure adjusted) with SLB (Table II). In the association population, significant correlations were identified between flecking and resistance to Fusarium ear rot and to MCB, although the correlations with MCB became nonsignificant after adjustment for population structure. The correlation coefficients were consistently higher in the association panel than in the NAM population. Nevertheless, significant correlations on a single NAM family basis were sometimes comparable to estimates in the association panel (Supplemental Table S2). Among the 25 NAM families analyzed independently, seven, 10, eight, and five families showed significant correlations between flecking and HR, NLB, GLS, and SLB, respectively, while the correlations with the other families were not significant (Supplemental Table S2). Of these significant correlations, all were positive (indicating an association between increased flecking and increased resistance or defense response) except for the NC350 family for GLS and for four of the five estimates for SLB, which were negative. Plant Physiol. Vol. 172, 2016

The strength of the correlations was generally highest for the more biotrophic (rust and NLB) and weakest for the more necrotophic (SLB and Fusarium ear rot) pathogens. The correlation between flecking and total kernel weight was only significant for the adjusted NAM values, where there was a very low negative correlation. Correlations between flecking and total kernel weight in the individual NAM families were significant in eight of 25 cases (Table II). In seven of these eight cases, negative correlations were observed, indicating a relationship between increased flecking and reduced kernel weight. To explore this trend, the average trait values of NAM lines grouped based on their flecking phenotypic classes were plotted against their total kernel weight values as well as their disease scores (Fig. 6). The trend in the plot suggested that total kernel weight was compromised when flecking scores were greater than 4. For GLS and NLB, there was a steady increase in disease resistance (lower disease scores) with increasing flecking severity. Increased SLB resistance was not observed until flecking score exceeded 5 (Fig. 6). DISCUSSION

Flecking, the mild-lesion phenotype observed on the leaves of some inbred maize lines, also has been referred to as genetic or physiological spotting in several other plant species (Ferdinandsen and Winge, 1930; Nair and Tomar, 2001; Makepeace et al., 2007). It is 1791

1792

bp

Positiona

0.38 0.27 0.80 0.31 0.43

9.39 11.82 6.02 5.10 5.22

9 127,467,238 10 113,149,572 Association population 5 17,321,110 7 127,583,397 7 155,619,901 SF SF –

JF, SF JF, SF

JF, SF JF, SF JF, SF

SF JF – JF, SF SF

SF SF JF, SF SF SF – SF –

QTL

GRMZM2G023347 GRMZM2G145152 GRMZM2G158062

GRMZM2G119067 GRMZM2G443509

GRMZM2G055238 RMZM2G026927 GRMZM2G128995

GRMZM2G132169 GRMZM2G097297 GRMZM2G144188 GRMZM2G049021 GRMZM2G179768

GRMZM2G172322 GRMZM2G161004 GRMZM2G380227 GRMZM2G043191 GRMZM2G109056 GRMZM2G329181 GRMZM2G077227 GRMZM2G178102

Gene Identifierd

ABI3-interacting protein3/prefoldin subunit Tsi1-interacting protein TSIP1 (DnaJ/Hsp40) Eukaryotic translation initiation factor-related

Glutathione reductase Splicing factor suppressor of abi3-5 Ser/Thr protein kinase (SIK, salt-inducible kinase) Inositol polyphosphate 5-phosphatase11 Lipoxygenase4 (LOX4) FASCICLIN-like arabinogalactan protein17 IQ calmodulin-binding motif domain 6-containing protein Homeobox-Leu zipper family protein/lipid-binding START domain-containing protein Laccase12 O-Methyltransferase family protein Dof-type zinc finger DNA-binding family protein Isochorismatase family protein Zinc-dependent matrix metalloproteinase/predicted glycosylphosphatidylinositol-anchored protein Ureide permease5 Ovate family protein4 (OFP4) Vacuolar ATP synthase subunit C (VATC)/V-ATPase C subunit/vacuolar proton pump C subunit (DET3) Cytochrome P450 (CYP75B1, D501, TT7) Protein phosphatase2C

Maize Gene Name or Best-Matching Orthologe

cwr, dfr cwr, dfr str, tf dfr, cwr dfr, vt aat, dfr tf, cwr vt, dfr rhx, dfr sig, str chp, vt dfr, str, chp pcd, str

4 3 102164 8 3 10234 0 3 3 10261 2 3 102179 1 3 10260 5 3 10219 5 3 10223

3 3 3 3

rhx, dfr rna, dfr str, sig aut, vt, dfr, sig rhx, dfr cwr sig, dfr tf

Functional Annotationg

0 10257 10217 10284 10274 5 1 2 3

0 9 3 10253 2 3 10241 5 3 102127 0 9 3 102135 6 3 102101 0

e Valuef

a

b c Physical map position of SNPs based on B73 maize reference genome assembly version 2. 2Log10 of the P value from GWA analysis based the full data set. RMIP based on d e f 100 subagging. Gene identity based on the B73 maize reference genome assembly version 2. Best Arabidopsis matching ortholog. e value based on protein sequence alignment g Key: aat, amino acid/ureide transporter; aut, autophagy; chp, molecular chaperone; cwr, cell wall related; dfr, defense response; pcd, with the best Arabidopsis matching ortholog. programmed cell death; rna, RNA binding and processing; rxh, redox homeostasis; sig, signaling; str, stress response; tf, transcription factor; tr, translational regulation; vt, vesicle trafficking/ remodeling.

0.3 0.54 0.39

14.01 25.43 19.82

9,664,459 153,836,827 77,776,660

0.27 0.26 0.35 0.96 0.37

6 6 8

15.14 16.10 13.04 27.86 11.52

0.26 0.52 0.53 0.45 0.38 0.38 0.28 0.32

RMIPc

183,645,872 192,900,213 2,343,988 133,587,094 205,924,654

11.67 16.18 15.18 17.30 11.81 10.37 14.58 12.26

2Log10 P Valueb

3 4 5 5 5

NAM population 1 13,093,772 1 188,173,260 1 197,541,610 1 253,792,996 1 264,209,352 1 286,021,810 3 5,994,780 3 123,263,674

Chromosome

The predicted functions and pathways of the maize candidate genes are based on the best-matching orthologs. A dash in the QTL column indicates that the SNP does not colocalize with a QTL.

Table I. SNPs significantly associated with the flecking phenotype identified with GWA analysis in the NAM and association populations

Olukolu et al.

Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

Figure 4. Flecking severity on the seventh leaf of inbred plants at 4 weeks after planting after low- and high-light exposure.

often considered to be a sign of a constitutive low-level defense response that might confer broad-spectrum resistance to diseases. Despite its potential utility as a convenient marker for disease resistance, there have been few attempts to understand its genetic basis and its relationship to disease resistance. A recent study using 300 lines from the maize intermated B73 3 Mo17 advanced intercross line mapping population identified low but moderately significant positive correlations between increased flecking and increased disease resistance and defense response (Vontimitta et al., 2015). In this study, using much larger populations, we show that the genetic architecture controlling variation in the flecking phenotype is complex. Despite the impact of environmental factors, such as light, on flecking (Fig. 4), the high heritability estimates reveal a strong genetic component underlying the expression of this trait. We present evidence that suggests that perturbation of a number of disparate pathways may affect the flecking phenotype and that, to some extent, a stronger visible flecking phenotype is correlated with higher general (broad-spectrum) disease resistance and defense response. While we cannot entirely discount the possibility that some of the flecking may have been due to low-level foliar infections, we are confident that most of the flecking observed was genetically controlled and not due to disease. Reasons for this include the fact that most of the lines showed no flecking at all (indicating low incidence of disease in the trial) and the distinct phenotypes that allow us to discriminate flecking from foliar diseases commonly endemic in field trial environments. Correlations between Flecking and Resistance to Disease and Pests, Severity of the HR Defense Response, and Total Kernel Weight

In the association and NAM populations, correlation coefficients between flecking and all other traits (disease and pest resistance and HR) were mostly positive Plant Physiol. Vol. 172, 2016

in direction (Table II). In other words, in most cases where a significant correlation was detected, stronger flecking was associated with enhanced disease and pest resistance or a stronger defense response. Within the NAM subfamilies, correlation levels were quite variable. However, the majority of significant correlations between flecking and the disease-related traits were again positive, except for one family for GLS and four of the five significant correlations for SLB (Supplemental Table S2). The consistently significant correlations between stronger HR and stronger flecking suggest that these two traits share some physiological basis. The five diseases evaluated range from the completely biotrophic common rust to the hemibiotrophic NLB, which can form intimate associations with living cells before killing them (Knox-Davies, 1974), and from the more necrotrophic GLS, which nevertheless has an extended nonsymptomatic invasive phase of about 2 weeks (Beckman and Payne, 1982), to SLB, which can kill host cells within about 24 h of infection (Jennings and Ullstrup, 1957). These generally higher positive correlations of flecking with rust, NLB, and GLS resistance traits compared with SLB suggest that flecking is more strongly associated with resistance to pathogens with biotrophic and hemibiotrophic rather than necrotrophic modes of infection. This is consistent with our evidence that the flecking and HR traits are associated and the fact that the HR primarily confers resistance against biotrophic pathogens. Two of our previous studies also showed that variation in HR severity was more correlated with resistance to pathogens with longer biotrophic phases (Olukolu et al., 2013, 2014). The lower correlations between flecking and Fusarium ear rot may be due to the specific mechanisms required to provide resistance to an ear rot rather than a foliar disease. Table II shows the correlations both with and without adjusting for the population structure (adjusted and 1793

Olukolu et al.

Figure 5. Total ROS content in 2-, 3-, and 4-week-old maize plants of several inbred lines. Inbred line mean values (averaged over values at 2, 3, and 4 weeks) that do not share common letters are significantly different (Duncan’s multiple range test). SE values are shown in error bars. RFU, Relative fluorescence units.

unadjusted scores, respectively). It is clear that population structure accounts for some of the correlation, although in most cases, the adjusted correlations remain significant. The main exception to this are the correlations between flecking and MCB in the association population, which appear to be largely due to population structure. Much evidence exists to suggest that constitutive expression of the defense response pathway can lead to a yield penalty (Brown, 2002). By plotting the average total kernel weight of lines grouped by their flecking scores, we observed an association between moderate flecking (score of 4 or greater) but not between mild flecking (score of 0.5–4) and reduced total kernel weight (Fig. 6) and total kernel weight. It should be noted that the total kernel weight data in this case were derived from a different experiment using the NAM population in two environments. Furthermore, although hybrid total kernel weight is more relevant to the commercial grower, total kernel weight here was measured in inbreds. With these caveats, we propose that low to mild levels of flecking might be a useful indicator of lines in which the disease resistance benefits of some of the processes that cause flecking (i.e. inducing the defense response) might be apparent without causing yield losses. JF Linkage and SF QTL Analysis in the NAM Population

JF linkage analysis in the NAM population detected 17 QTLs with mostly small effects (Supplemental 1794

Files S5 and S6). QTL analyses in the SF identified some QTLs with as much as 17.7% partial contribution to phenotypic variance (Supplemental File S6). Between two and seven QTLs were detected in each SF analysis. No significant epistatic QTL interactions were detected in this study. Other studies of quantitative traits in the maize NAM population have similarly failed to find significant epistasis (Kump et al., 2011; Poland et al., 2011; Olukolu et al., 2014). We should note that the structure of the NAM population, with 26 alleles segregating at each locus, has limited power to detect moderate-effect epistatic interactions between any two alleles. Fourteen of the 17 flecking QTLs overlapped with previously identified QTLs associated with HR severity due to Rp1-D21 (Olukolu et al., 2014) and resistance QTLs underlying SLB (Kump et al., 2011), NLB (Poland et al., 2011), and GLS (Benson et al., 2015; Supplemental Fig. S3). Candidate Genes and Pathways

For each associated SNP identified from the NAM and association populations, the gene harboring the SNP or the closest annotated gene in the sequenced B73 maize genome (Schnable et al., 2009) was identified as the candidate gene. The rapid breakdown of linkage disequilibrium in the maize NAM and association populations (Olukolu et al., 2013), the high-density genotypic data available, and the high statistical Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

Table II. Spearman phenotypic correlation coefficients between flecking and other traits, which include the HR defense response, disease scores (SLB, NLB, GLS, Fusarium ear rot, and common rust), and resistance to the MCB (scored as kernel resistance and tunnel length in the stem) Phenotypic values were adjusted for population structure in both association and NAM populations. HR, Rp1-D21-induced HR measured as mutant to wild-type height ratio (Olukolu et al., 2014); KR, kernel resistance (recorded at harvest as the damage on the main ear of the five infested); TL, tunnel length (mean length of stem tunnels made by borers on the five infested plants, which were longitudinally split at the time of harvest); TKW, total kernel weight (Hung et al., 2012; the subscript inv implies that the original rating was rescaled to reflect the score in same direction of phenotype [i.e. the value corresponding to the high score became the low score and vice versa]). The significance of correlation coefficients (r) is shown with asterisks: ****, P , 0.0001; ***, P , 0.001; **, P , 0.01; and *, P , 0.05 (detailed estimates for individual NAM populations shown in Supplemental Tables S3–S5). A positive correlation implies that higher flecking is associated with higher disease resistance or HR defense response. The values of nonsignificant correlations also are shown, and in these cases, the P value of the correlation is indicated. NA indicates that the population was not assessed for both traits, so a correlation could not be calculated. The association panel scoring scale is as follows: flecks, 0 (no flecking) to 10 (severe); HR, 0 (severe) to 1 (mild); rust, 0 (resistant) to 10 (susceptible); NLB, 0% (susceptible) to 100% (resistant); GLS, 0% (susceptible) to 100% (resistant); SLB, 0 (susceptible) to 9 (resistant); Fusarium, 0 (susceptible) to 10 (resistant); and KR, 1 (susceptible) to 9 (resistant). The NAM population scoring scale is as follows: flecks, 0 (no flecking) to 10 (severe); HR, 0 (severe) to 10 (mild); SLB, 0 (resistant) to 9 (susceptible); NLB, 0% (resistant) to 100% (susceptible); and GLS, 0% (resistant) to 100% (susceptible). ns, Not significant. Population

Population Structure

HR

Disease Scores Rustinv

NLB

Association Unadjusted

NAM (JF)

GLS

MCB SLB

0.33**** 0.36**** 0.38**** 0.34**** 0.30**** (n = 231) (n = 244) (n = 272) (n = 267) (n = 278) Adjusted 0.24*** 0.31**** 0.22*** 0.17** 0.16** (n = 231) (n = 244) (n = 272) (n = 267) (n = 278) Unadjusted 0.08**** NA 0.24**** 0.21**** 0.05*** (n = 3,354) (n = 4,257) (n = 3,474) (n = 4,770) Adjusted 0.11**** NA 0.14**** 0.08**** 20.07**** (n = 3,354) (n = 4,257) (n = 3,474) (n = 4,770)

power afforded by these populations suggest that many of the highly associated SNPs will be very close to the causal polymorphisms, such that the candidate gene selected will frequently be the causal gene. A recent publication showing that associations identified in the NAM population with a variety of traits are enriched within and around genes, and specifically regulatory genes, supports these assumptions (Wallace et al., 2014). Based on functional annotations of bestmatching gene orthologs, 14 of the 22 candidate genes have been directly implicated in the defense response (Table I). The predicted roles that these genes may play in modulating pathways associated with flecking, including the defense response, redox homeostasis, and light response, are discussed briefly below. Cell Wall Related

The plant cell wall plays a crucial role during plantpathogen interaction and during HR. Plant cell wallrelated defense mechanisms include (1) recognition of pathogens by signaling receptor molecules such as Leu-rich repeat receptor-like kinases; (2) protection from pathogenic penetration by inhibiting cell wall-degrading enzymes, structural and chemical remodeling of breached cell wall sites, and preventing pathogen growth with antimicrobial proteins and secondary metabolites; (3) structural and chemical strengthening of the cell wall for penetration resistance as well as localizing host cellular damage due to ROS during HR; and (4) trafficking of defense compounds (Hückelhoven, 2007; Malinovsky et al., 2014). The five candidate genes from this study predicted to be Plant Physiol. Vol. 172, 2016

Fusariuminv

KR

0.21*** 0.23*** (n = 267) (n = 265) 0.12* ns (n = 267) (n = 265) NA NA NA

NA

TLinv

0.17** (n = 267) ns (n = 267) NA NA

TKW

ns (n = 260) ns (n = 260) ns (n = 4,766) 20.06**** (n = 4,766)

involved in cell wall modification are FASCICLIN-LIKE ARABINOGALACTAN PROTEIN17 (FLA17), OVATE FAMILY PROTEIN4 (OFP4), LACCASE12 and S-adenosylL -Met-dependent O-methyltransferases (OMTs), and an isochorismatase family protein (Table I). Fasciclin domains of the FLAs are ancient cell adhesion domains conserved across various phyla. Often found in the extracellular matrix, their role in cell-tocell interactions and communication provides a basis for structural, positional, and environmental signals during plant development. The biological function of FLAs in higher plants is not well clarified. Nevertheless, studies suggest a role in cell expansion and cellulose synthesis in the secondary cell wall (MacMillan et al., 2010, 2015). Homologs of OFP4 are transcriptional repressors in the plant-specific ovate family that regulate plant growth and development. The OFP4 transcription factor interacts with KNAT7 to regulate secondary cell wall formation (Li et al., 2011; Wang et al., 2011). Lignin biosynthesis has been shown to play a key role in cell wall apposition-mediated defense against fungal penetration. Laccases from plant cell suspension cultures polymerize lignin monomers in vitro (Sterjiades et al., 1992; Bao et al., 1993). Laccases also have been localized to the cell wall of lignifying sycamore (Acer pseudoplatanus) cells (Driouich et al., 1992). OMTs or caffeic acid 3-O-methyltransferases (CAOMTs) catalyze the methylation of hydroxycinnamic acid derivatives (caffeic acid and 5-hydroxyferulic acid) to synthesize lignin. Silencing of CAOMT produced increased penetration efficiency by Blumeria graminis f. sp. tritici in wheat (Bhuiyan et al., 2009). 1795

Olukolu et al.

Figure 6. Total kernel weight and SLB, GLS, and NLB disease scores of NAM RILs in flecking phenotypic classes. Mean values that do not share common letters are significantly different (Duncan’s multiple range test). On the primary y axis, disease severity scores were standardized by rescaling scores from 0 (resistant) to 10 (susceptible). TKW on secondary y axis is measured in grams. SE values are shown in error bars.

Salicylic Acid- and Jasmonic Acid-Dependent Response Pathways

The salicylic acid (SA)- and jasmonic acid (JA)dependent response pathways are thought to act antagonistically during the response to biotrophic and necrotrophic pathogens, respectively. Several candidate genes identified here might be involved in modulating SA/JA levels (Table I). Alteration of lignin levels can affect SA levels in complex ways. The biosynthesis of both SA and lignin utilizes Phe via the phenylpropanoid pathway. Plants also employ an alternative route for SA biosynthesis that may involve the isochorismatase candidate gene identified in this study (Chen et al., 2009; Gallego-Giraldo et al., 2011). The SAdependent defense response pathway induces the phenylpropanoid pathway. The LOX4 candidate gene has been shown to be induced in leaves by JA and also highly up-regulated in resistant maize genotypes upon infection by Fusarium ear rot (Maschietto et al., 2015). Tsip1, an SA-responsive gene, is important for responses to both biotic and abiotic stress. Transgenic 1796

expression of Tsip1 confers pathogen resistance and salt tolerance (Ham et al., 2006). Redox Homeostasis and PCD

ROS are key modulators of PCD processes, including HR (Gechev et al., 2004), and transient oxidative bursts due to biotic and abiotic stress can lead to high concentrations of ROS, which can be very damaging to the cell. Consequently, plants have complex systems that regulate redox homeostasis. Four of the 22 candidate genes, glutathione reductase, a cytochrome P450, LOX4, and a eukaryotic translation initiation factor-related gene, are potentially involved in PCD and/or redox homeostasis (Table I). Besides reducing hydrogen peroxide, a highly damaging reactive species, by reducing the oxidized form of glutathione peroxidase, glutathione reductase also removes xenobiotics and acts as a cofactor in some detoxifying enzymes. Mutants with altered contents of glutathione compromise the oxidative balance of cells and have been shown to be hypersensitive to stress Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

(Creissen et al., 1999; Mittler, 2002). Cytochrome P450s are a large superfamily of proteins that often function to deliver one or more electrons to reduce iron and eventually molecular oxygen (Schuler, 1996). Several studies have shown that eukaryotic translation initiation factor-related proteins participate in the early response during PCD by modulating the rate of protein synthesis through changes in the phosphorylation of translation initiation factor polypeptides and the association of these factors into functional complexes or targeted cleavage of factors by cellular proteases (Morley et al., 2005). Vesicle Trafficking and Remodeling

Vesicle trafficking is important for the defense response in two major ways: (1) to reposition organelles at the pathogen penetration site, and (2) the secretion of antimicrobial compounds. The aggregation of organelles beneath the pathogen penetration site, probably due to cytoskeleton remodeling, establishes a physical barrier by increasing the density of cellular components (Frei dit Frey and Robatzek, 2009). The candidate genes predicted to be involved in vesicle trafficking and remodeling include a zinc-dependent matrix metalloproteinase/predicted glycosylphosphatidylinositolanchored protein, vacuolar ATP synthase subunit C, and ABI3-INTERACTING PROTEIN3/prefoldin (Table I). Matrix metalloproteinases (MMPs) are a family of zinc-dependent endopeptidases that function in physiological and pathological processes. Treatment of tomato (Solanum lycopersicum) plants with defenserelated hormones (SA, JA, and ethylene) or challenge with pathogens resulted in the up-regulation of MMP gene expression, while a knockdown of MMP genes resulted in reduced resistance (Li et al., 2015). V-ATPase, a multimeric complex, plays a role in establishing an acidic pH that provides optimal conditions for enzymes in lumens of cellular compartments. Besides establishing an electrochemical proton gradient, which in turn drives many secondary transport processes, its subunits also have direct roles in the regulation of vesicular trafficking along both exocytotic and endocytotic pathways (Marshansky and Futai, 2008). Molecular chaperones interact with nascent polypeptide chains to facilitate proper folding. Chaperones have been reported to be involved in the plant defense response and in the mediation of redox stress (Papp et al., 2003). ABI3-INTERACTING PROTEIN3/ prefoldin is a heterohexameric chaperone required for the proper functioning of actin and tubulin-based cytoskeleton and, consequently, is important for vesicle trafficking (Vainberg et al., 1998). Signal Transduction

Several lipids and lipid-related molecules have been implicated in signaling and defense response to Plant Physiol. Vol. 172, 2016

biotrophic and necrotrophic pathogens (Shah, 2005). Four of the candidate genes are potentially involved in signaling (Table I). The candidate gene, INOSITOL POLYPHOSPHATE 5-PHOSPHATASE11, functions within the phosphoinositide-mediated signaling pathway, which is required for a robust plant defense response for both basal defense and systemic acquired resistance response (Hung et al., 2014). Protein phosphatase 2C genes are implicated as negative regulators of protein kinase cascades activated as a result of stress. They have been associated with abscisic acid signaling and wound signaling pathways, which are associated with the JA-dependent defense response (Meskiene et al., 2003; Schweighofer et al., 2007). Other candidate genes involved in signal transduction include a Ser/Thr protein kinase (SALT-INDUCIBLE KINASE1), which plays a role in modulating signaling networks that correct disruptions in cellular ion balance (Bertorello and Zhu, 2009), and an IQ calmodulinbinding motif domain 6-containing protein, which plays roles in calcium signaling and other biological processes (Bähler and Rhoads, 2002). Alternative Splicing

The splicing SUPPRESSOR OF ABI3-5 gene is important in the regulation of plant immunity mediated by the RLKs SUPPRESSOR OF NPR1-1, CONSTITUTIVE4, and CHITIN ELICITOR RECEPTOR KINASE1 (Zhang et al., 2014). Transcription Factor

Transcription factors regulate plant stress responses in multiple and complex overlapping signaling pathways. Three of the candidate genes are predicted to be transcription factors (Table I). The Dof motif in the Dof-type zinc finger gene is an evolutionarily conversed transcription factor that can function as a transcriptional activator or repressor of tissue-specific and light-regulated gene expression in maize (Yanagisawa and Sheen, 1998). Homeobox-Leu zipper family protein/lipid-binding START domain-containing proteins are conserved transcription factors unique to plants and have been linked to redox homeostasis, light signaling, and disease resistance (Elhiti and Stasolla, 2009; Fu et al., 2009). Role of Light and Redox Homeostasis in Flecking Severity

ROS generated during electron transport, which accompanies oxidative phosphorylation and photosynthesis, lead to oxidative stress and extensive DNA damage (Bhattacharjee, 2012). Les mutants in several species require light for the initiation and/or propagation of lesions (Buckner et al., 2000; Negeri et al., 2013). We show here that flecking, at least in some lines, also is light responsive (Fig. 4) The relationship between ROS production and accumulation can be complex due to the differential 1797

Olukolu et al.

availability of antioxidants. Some lines might show more accumulation than others following ROS production, due to the differential levels of cellular damage and the efficiency of antioxidant function (Noctor et al., 2007). The higher levels of ROS observed in mild flecking lines compared with lines in which no flecking was observed (Fig. 5) suggest that ROS may be associated with flecking. The lower levels of ROS observed in lines with the highest levels of flecking may be due to the cell death caused by the accumulation of high levels of ROS. CONCLUSION

In this study, we have provided a detailed analysis of the genetic architecture controlling foliar flecking in maize, a commonly observed but little-studied phenotype. We provide evidence linking this phenotype to disease resistance and the defense response. It is important to note that the correlations between flecking and disease resistance and the defense response, although significant, are moderate to low in magnitude. It is likely that multiple pathways and processes contribute to leaf flecking and that some, although not all, of them also are related to plant defense. In particular, we provide evidence linking flecking to processes controlling cell wall modification, SA- and JA-dependent pathways, stress response, redox homeostasis, and vesicle trafficking/remodeling. Attempts to engineer plants for broad-spectrum disease resistance via constitutive expression of the defense response have often resulted in compromised plant growth and significant yield loses (Agrawal and Karban, 1999; Yin et al., 2000; Maleck et al., 2002). We present data to suggest that mild flecking might be used as a marker for increased broad-spectrum disease resistance that does not compromise yield.

replication); a greenhouse in 2011 (GH11; one replication); and field trials in Homestead, Florida, over the winter season of 2011/2012 (FL11/12; two replications) and in Clayton, North Carolina, during the summers of 2012 and 2013 (NC12 and NC13; one replication in 2012 and two replications in 2013). A randomized complete block design was used in FL11/12 and NC13, while an augmented randomized incomplete block design was used in GC11, GH11, and NC12. In the field trials, a border row of a constant genotype was planted around the trials to eradicate border row effects. Standard fertilizer, herbicide, and pesticide applications were used to ensure normal plant growth during the trial. Overhead irrigation was applied as required. Ten and six kernels of each line in the association and NAM populations, respectively, were sown in 2-m rows. Spacing between rows was 0.97 m, while the alley at the end of each plot was 0.6 m. For experiments conducted in the greenhouse, plants were grown in pots filled with a 3:1 mixture of Farfard (Sun Gro Horticulture) 4P mix (50% peat, processed pine bark, perlite, and vermiculite) and sterilized soil. A slow-release fertilizer, Osmocote 19-6-12, also was included in the soil mix, while plants were watered twice daily. The plants were grown continuously at about 26°C during the day and about 22°C at night. The actual temperature rarely deviated more than 4°C from the set target temperatures mentioned above. Supplemental lighting at 14 h per day was employed to maintain an approximately 16-h daylength. For experiments conducted in the growth chamber, pots were filled with a mix of sterilized soil and gravel. Plants were watered twice daily with deionized water in the morning and then a nutrient solution in the evening. Chamber temperature was maintained at 30°C during the day and 20°C at night. Lighting was maintained at 12 h per day through the study.

Evaluation of the Flecking Phenotype In all environments and populations, severity scores of the flecking phenotype were assigned based on a scale of 0 to 10, with 0 indicating no lesion and 10 indicating a completely dead plant due to necrotic lesions. This rating scale was developed for a more severe lesion phenotype (Chintamanani et al., 2010) based on an autoactive HR. Since the flecking phenotype we are scoring here is a more moderate phenotype, 6 was the highest score assigned. Each trial was scored for flecking three times. In the growth chamber and greenhouse environments, scoring was performed at 2, 4, and 6 weeks after planting. For the other trials performed in field environments, scoring was performed at 4, 8, and 12 weeks after planting. The sAUC, a widely accepted statistic in plant pathology, where it is known as the standardized area under the disease progress curve, was used to estimate flecking severity throughout each trial from the three scores (Shaner and Finney, 1977). The sAUC was estimated as follows: the average value of two consecutive ratings was calculated and multiplied by the number of days between the ratings. The summation of values over all intervals was then divided by the total number of days over which evaluations were performed to determine the weighted average.

MATERIALS AND METHODS Mapping Populations Generation of the maize (Zea mays) NAM population has been described previously (Buckler et al., 2009; McMullen et al., 2009). It consists of 25 biparental populations of 200 RILs each with inbred B73 as a common parent. The association population consists of 302 inbred lines (of which 279 are used in this study) that represent the global diversity available in public-sector cornbreeding programs. Pedigrees of these lines also have been described previously (Gerdes and Tracy, 1993; http://www.ars-grin.gov/).

Field, Greenhouse, and Growth Chamber Trials The 4,998 RILs of the NAM population were evaluated twice in 2012 and 2013 at Clayton, North Carolina (NC12 and NC13). A single replication of the NAM RILs was planted in each year. To provide an estimate of experimental error, an augmented randomized incomplete block design was employed with the founder parental inbred lines used as repeated checks. Each NAM family was allocated to a block, and then each family block was divided into 10 equal subblocks. The two parents corresponding to each NAM family were used as repeated checks in each of the 10 subblocks within each NAM family block for both years. The 279 inbred lines of the association population were evaluated for the flecking phenotype in five environments: a growth chamber in 2011 (GC11; one 1798

Disease and Pest Resistance, Rp1-D21-Associated HR, and Total Kernel Weight Scores Disease scores for SLB (causal agent, Cochliobolus heterostrophus), GLS (causal agent, Cercospora zeae-maydis), NLB (causal agent, Exserohilum turcicum), and Fusarium ear rot (causal agent, Fusarium verticillioides) were obtained from previous studies in the association population (Wisser et al., 2011) and NAM population (Kump et al., 2011; Poland et al., 2011; Benson, et al., 2015). Maize common rust scores and pest resistance scores (kernel resistance and tunnel length) for the MCB (causal agent, Sesamia nonagrioides) were obtained only for the association population (Samayoa et al., 2015; Olukolu et al., 2016). Rp1-D21 HR scores (mutant to wild-type height ratio) were obtained from previous studies in the association population (Olukolu et al., 2013) and NAM population (Olukolu et al., 2014). Data on total kernel weight were taken from the total kernel weight of two plants, scored across 11 environments, obtained from a previous study in the NAM and association populations (Hung et al., 2012).

Statistical Analyses in the NAM Population Phenotypic Data To obtain heritability estimates, data were analyzed with a strategy described by Piepho et al. (2006) using Proc Mixed in SAS version 9.3 (SAS Institute, Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

2000-2004). A mixed linear model, with RILs, environment, block (environment), and RILs-by-environment as random effects, was developed while creating an additional random variable that incorporated values for observations in the controls (i.e. parental founders used as repeated checks). LSmean values for RILs were estimated based on this mixed model but with RILs as a fixed effect. Correlation analyses were performed based on the LSmean values using Proc Corr in SAS version 9.3 (SAS Institute, 2000-2004). Since the flecking data were not distributed normally, we used the Spearman rank correlation, which does not require the assumption about data normality, as does the Pearson correlation. To perform the Spearman rank correlation analysis while accounting for the effect of the NAM population structure, each NAM RIL LSmean value was adjusted for its population mean. The residual diagnostics from the Proc Mixed model revealed that the flecking data were not distributed normally. To adjust for environmental and other nongenetic effects, data were analyzed to estimate BLUP values in a generalized linear mixed model using Proc GLIMMIX in SAS version 9.3 (SAS Institute, 2000-2004). Since flecking scores approximate count data, a Poisson distribution was specified in the generalized linear mixed model, with natural logarithm as the link function. An overdispersion parameter was included in the model. The ratio of generalized x 2 statistic with the residual degrees of freedom was examined (less than 1) as the usual residual dispersion estimate to protect against the overdispersion problem during model fitting. Consequently, BLUPs of the line random effect were used as input phenotype data for QTL and GWA analyses.

GWA Analysis A total of 4,521 NAM lines for which we had both genotypic and phenotypic data were used in GWA analysis. Models were fit for each of the maize chromosomes one at a time. RIL residual values from the final fitted QTL model on each chromosome, excluding all markers on the same chromosome under consideration, were estimated in SAS software version 9.3 (SAS Institute, 20002004). GWA was performed following a forward selection regression of 26.5 million HapMap version 2 SNPs to subsamples of the flecking phenotypic data (residual values) on one chromosome at a time. To calculate RMIP values, the subsampling or subagging method was employed during the GWA analysis (Valdar et al., 2009). Forward selection regression was implemented 100 times based on a subsampled data set. Each of the subsampled data sets constituted 80% of RILs from each NAM family (Kump et al., 2011; Tian et al., 2011). The population main effect was incorporated into the model before the addition of marker terms during forward selection. Only markers determined as significant at P , 1 3 1026 (Panagiotou and Ioannidis, 2012) were fitted into the GWA analysis model. RMIP values (between 0 and 1) were obtained for each SNP marker and represent the proportion of subsamples in which the SNP marker was selected. An RMIP threshold of 0.25, established based on simulations by Valdar et al. (2009), was used subsequently to report the most robustly significant associations.

Statistical Analyses in the Association Population

Genotypic Data, SNP Imputation, and SNP Projection A total of 7,386 SNP markers scored on all 4,998 RILs were used for genetic linkage and QTL analyses. The development of genotyping-by-sequencing (GBS) SNP markers, imputation of missing data in the NAM founder parents, and SNP projection from founder parental genotypes on the RILs have been described previously (Elshire et al., 2011; Romay et al., 2013; Olukolu et al., 2014).

SF and JF Linkage Analysis Both genotyping and the construction of the NAM linkage map were implemented by McMullen et al. (2009), while the joint linkage mapping has been described previously (Buckler et al., 2009). The SF QTL analysis of all 25 populations was performed independently based on composite interval mapping in Windows QTL Cartographer version 2.5 (Silva et al., 2012). Log of the odds (LOD) thresholds of approximately 3 were determined after 1,000 permutation tests with an a set at 0.05. Linkage and QTL maps were prepared using the MapChart software version 2.2 (Voorrips, 2002). Prior to joint stepwise regression analysis in PROC GLMSELECT, as implemented in SAS software version 9.3 (SAS Institute, 2000-2004), LOD thresholds were determined after 1,000 permutation tests with an a set at 0.05. The model included family main effects and single marker effects nested within families. After initial detection of QTLs, the stepwise regression model was optimized based on an iterative process by sequentially dropping SNPs from the model, testing the fit of adjacent SNPs until the eighth SNP marker (i.e. 1.6 cM away) from the originally selected SNP, and then fitting the best SNP marker in the QTL region back into the model. Allele effects at each SNP marker in the final model were determined simultaneously using the solution option of Proc GLM in SAS software version 9.3 (SAS Institute, 2000-2004). The QTL support intervals were determined in SAS software version 9.3 by including a flanking SNP marker to the full model one at a time from the QTL peak at a step of 0.2 cM. This was implemented at the left and right sides of the QTL peak. The boundary of the support interval was the last marker at which the QTL was significant at P = 0.05. To test for the significance of digenic epistatic interactions, a subset of 1,409 SNPs were obtained from the available 7,386 SNPs at uniform 1-cM marker intervals. All pairwise combinations of the 1,409 markers were tested separately with a model that included population main effects, two marker main effects nested in populations, and the marker-marker interaction nested in populations by extending the method of Holland (1998). The analysis was performed across the 25 NAM families, while the QTL LOD threshold was estimated following a permutation test at a critical value with a , 0.05 (1,000 permutations). Marker pair interactions with P values less than the permutation test-based threshold were included in the final joint linkage model without regard to the significance of the main effects of the markers. Each such pair of markers and their interaction were added to the final additive joint linkage model one at a time; the P value of the interaction in this full model was used to determine if the epistatic interaction improved the model fit. Plant Physiol. Vol. 172, 2016

Phenotype Modeling To estimate heritability values, a mixed model was implemented with inbreds, environment, and inbred-by-environment as random effects. To estimate inbred LSmean values adjusted for environmental effects, data were analyzed with a mixed model fitted with inbreds as fixed effects and environment and inbred-by-environment as random effects using Proc Mixed in SAS version 9.3 (SAS Institute, 2000-2004). The residual diagnostics from the Proc Mixed model revealed that the data are not distributed normally. A similar transformation procedure (described above to estimate BLUP values) employed for the transformation of NAM population flecking data was applied to the association population. Similarly, a Spearman rank correlation analysis was performed with the LSmean values using Proc Corr in SAS version 9.3 (SAS Institute, 20002004). To perform the Spearman rank correlation analysis while accounting for the effect of the population structure, each inbred line LSmean value was adjusted by regressing line values on covariates (Q1 and Q2) in derived population structure estimates of the probability that an inbred line belongs to any of three subpopulation groups in the diversity panel (Olukolu et al., 2013). The residual effect for each line from this regression was used as an adjusted value for the correlation analysis.

Genotypic Data The genotypic data, a total of 246,497 SNPs, consists of two SNP data sets generated from the Illumina maize 50k array (Ganal et al., 2011) and by a GBS approach (Romay et al., 2013). The Illumina maize array genotypes were filtered from 57,838 to 47,519 SNPs based on SNPs with a minor allele frequency value greater than 0.05 and less than 20% missing data (Olukolu et al., 2013). The same filter criteria were applied to the GBS SNP data set according to Zila et al. (2014) to obtain 200,978 SNPs from an original set of 681,257 SNPs.

GWA Analysis The confounding effects of population structure were accounted for during genome-wide association studies (GWAS) in order to avoid the systematic bias that produces false-positive associations (Hirschhorn and Daly, 2005). This was implemented by incorporating the kinship matrix (the genetic relationship between all pairs of lines) as a term in the mixed linear model. The kinship matrix was computed previously in this same set of inbred lines as described by Olukolu et al. (2013). The Tassel version 4.1.8 software (Bradbury et al., 2007) was used for GWAS based on a mixed linear model. The vector of phenotypes (y) was modeled as follows: y ¼ Xb þ Zu þ e; where b = a vector composed of fixed effects, including the SNPs being tested; u = a vector of the random additive genetic background effects associated with the inbred lines; e = a vector of residual effects; and X and Z are incidence 1799

Olukolu et al. matrices that relate y to b and y to u. The variances of the random effects are modeled as follows: VarðuÞ ¼ 2KVg ; where K = the n 3 n matrix of pairwise kinship coefficients that define the degree of genetic covariance between inbred lines and Vg = genetic variance (Yu et al., 2006). The Restricted Maximum Likelihood estimates of variance components were computed based on an efficient mixed-model association algorithm method (Kang et al., 2008; Zhang et al., 2010). To increase statistical power and computational speed, the optimum compression mixed linear model and P3D options were employed by clustering inbred lines into groups (Zhang et al., 2010). The P values for the 246,497 association tests were used to compute the positive false discovery rate associated with each P value observed using QVALUE version 1.0 software (Storey and Tibshirani, 2003), an R package (R Core Team, 2008). The Q value threshold was set at 0.3 (2log10 P = 5.1). P value thresholds also were computed based on a permutation test (a = 0.05), with the structure of the SNPs and inbred lines maintained while flecking scores were randomized based on 1,000 permutations using the R software. The permutation test-based P value threshold was 3.47 3 1026 (2log10 P = 5.46). Following GWA analysis on the full data set, a second round of GWA analysis was performed based on a subsampling (or subagging) method (Valdar et al., 2009). Subsampling was performed 100 times, with each of the subsampled data sets constituting 80% of the inbred lines. In addition to the P value threshold cutoff, only SNP markers detected in at least 25 of the 100 subsamples (i.e. RMIP threshold of 0.25) were considered significant (Panagiotou and Ioannidis, 2012). The Manhattan and QQ plots were generated using the qqman R package.

Candidate Gene Selection Genes that colocalized with (an associated SNP inside a gene) or that were immediately adjacent to the most highly associated SNP markers at each locus in the association and NAM populations were identified using the maize B73 reference genome version 2 obtainable from the MaizeGDB genome browser (Andorf et al., 2010) or the maizesequence genome browser (Schnable et al., 2009). Functional annotation of candidate genes was performed with the BLASTp (Altschul et al., 1997) and conserved domain search (Marchler-Bauer et al., 2005) tools on the National Center for Biotechnology Information Web site as well as annotations available at the MaizeGDB (http://gbrowse.maizegdb.org/gb2/gbrowse/maize_v2/). Additional annotation was achieved by review of the literature specific to each gene and domain.

Evaluating Flecking Severity under Variable Light Intensity To evaluate the influence of light intensity on flecking severity under greenhouse conditions and natural light, maize inbed lines were initially grown under a shade (gray wire mesh) until 4 weeks old (low light intensity). Five replicated plants were left under this shaded condition, while five replicated plants were transferred to supplemental lighting conditions at 14 h per day and without shade (high light intensity). Light intensity under both low (90–120 mE m 2 2 s 21 or 9–12 W m 2 2 or 900–1,200 lx) and high (280–300 mE m 22 s 2 1 or 28–30 W m 2 2 or 2,800–3,000 lx) intensity was measured using a LI-COR LI-185B Quantum/Radiometer/Photometer equipped with an LI-190SB quantum sensor. Flecking severity was scored subsequently on 5-week-old plants after 1 week of the differential light treatments.

Quantification of ROS The fifth leaf from select inbred lines grown in a greenhouse was sampled at 2, 3, and 4 weeks by taking leaf punches. Four biological replicates were obtained for each line. The samples were frozen immediately in liquid nitrogen and stored in a 280°C freezer. ROS content in the leaf sample was quantified using 29,79-dichlorodihydrofluorescein diacetate based on a protocol described previously by Jambunathan et al. (2010). A total of 30 mg of leaf tissue ground in liquid nitrogen was suspended in 0.5 mL of 10 mM Tris-HCl, pH 7.2, and centrifuged at 12,000g for 20 min at 4°C. The 1800

supernatant was tranfered to a fresh 1.5-mL tube and centrifuged again at 12,000g for 20 min at 4°C to obtain a clear solution. The concentration of ROS was determined using a flourometer (BioTek Synergy H1 Hybrid Plate Reader) following conversion of the nonfluorescent 29,79-dichlorofluorescein to the ROSoxidized highly fluorescent 29,79-dichlorofluorescein. Total protein content in each sample was determined based on the Bradford protein assay (Bradford, 1976) using the Bio-Rad protein assay reagent. The concentration of ROS was expressed in relative fluorescence units per mg of protein content.

Supplemental Data The following supplemental materials are available. Supplemental Figure S1. Distribution of phenotypic values for flecking across the NAM population shown in a bean plot. Supplemental Figure S2. Distribution of phenotypic values for flecking across the association population (three field trials) shown in a bean plot. Supplemental Figure S3. NAM joint-linkage QTL analysis for the flecking trait across all 10 maize chromosomes/linkage groups. Supplemental Figure S4. QQ plot showing the distribution of P values from GWAS in the association population. Supplemental Figure S5. Results of GWAS in the NAM population showing the SNPs with RMIP values above 0.05. Supplemental Figure S6. Results of GWAS in the association population showing the SNPs with RMIP values above 0.05. Supplemental Figure S7. Relative positions and effects of the eight colocalizing SNPs and QTL pairs. Supplemental Table S1. Spearman phenotypic correlation coefficients between flecking scores across five environments in the association population. Supplemental Table S2. Spearman phenotypic correlation coefficients in the individual subfamilies constituting the NAM population between flecking and other traits. Supplemental File S1. Phenotypic values for flecking, HR severity traits (Les and mutant to wild-type height ratio), disease scores (SLB, NLB, and GLS) and yield data in the NAM population. Supplemental File S2. Phenotypic values for flecking, HR severity traits (Les and mutant to wild-type height ratio), disease scores (SLB, NLB, GLS, and rust), flowering dates (days to anthesis), and total kernel weight data in the inbred association population. Supplemental File S3. Genotypes of the NAM RILs at 7,386 SNP loci. Supplemental File S4. Genetic linkage map of the NAM population used in this analysis. Supplemental File S5. QTL genetic and physical map support interval for flecking based on joint linkage analysis. Supplemental File S6. QTL peaks, LOD scores, allelic effects, and percentages of the contribution to phenotypic variance (r2; based on total adjusted phenotypic data [BLUPs]) of flecking QTLs computed for each of the 25 NAM populations. Supplemental File S7. P values, corresponding Q values, and r2 (calculated as the percentage contribution to phenotypic variance of total adjusted phenotypic data [BLUPs]) from GWAS on the full data set of the flecking phenotype in the association population Supplemental File S8. Significant SNPs at thresholds of –log10 P = 4.5 and RMIP = 0.05 for the flecking phenotype in the association population. Supplemental File S9. Significant SNPs at thresholds of –log10 P = 6 and RMIP = 0.05 for the flecking phenotype in the NAM population. Supplemental File S10. Table of NAM population GWAS candidate genes and functional annotations with references. Supplemental File S11. Table of association population GWAS candidate genes and functional annotations with references. Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

ACKNOWLEDGMENTS We thank Major Goodman and the Maize Genetics Cooperation Stock Center for donating seed and Cathy Herring and the staff of the Central Crops Research Station, especially David Rhyne and Greg Marshall, for help with field work; Ed Buckler, Peter Bradbury, Jeff Glaubitz, and the Maize Diversity project team for access to the GBS genotype data set and scripts required to run the joint linkage analysis and GWA analysis; Guri Johal for useful discussion; Toni Kazic for the gift of seeds of the Les mutant lines; and the MaizeGDB database (http:// www.maizegdb.org), which was essential to this work. The growth chamber facilities were provided by the North Carolina State University Phytotron. Received November 30, 2015; accepted September 22, 2016; published September 26, 2016.

LITERATURE CITED Agrawal AA, Karban R (1999) Why induced defenses may be favored over constitutive strategies in plants. In R Tollrian, CD Harvell, eds, The Ecology and Evolution of Inducible Defenses. Princeton University Press, Princeton, NJ, pp 45–61 Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25: 3389–3402 Andorf CM, Lawrence CJ, Harper LC, Schaeffer ML, Campbell DA, Sen TZ (2010) The Locus Lookup tool at MaizeGDB: identification of genomic regions in maize by integrating sequence information with physical and genetic maps. Bioinformatics 26: 434–436 Bähler M, Rhoads A (2002) Calmodulin signaling via the IQ motif. FEBS Lett 513: 107–113 Balint-Kurti PJ, Wisser R, Zwonitzer JC (2008) Use of an advanced intercross line population for precise mapping of quantitative trait loci for gray leaf spot resistance in maize. Crop Sci 48: 1696–1704 Balint-Kurti PJ, Yang J, Van Esbroeck G, Jung J, Smith ME (2010) Use of a maize advanced intercross line for mapping of qtl for northern leaf blight resistance and multiple disease resistance. Crop Sci 50: 458–466 Balint-Kurti PJ, Zwonitzer JC, Wisser RJ, Carson ML, Oropeza-Rosas MA, Holland JB, Szalma SJ (2007) Precise mapping of quantitative trait loci for resistance to southern leaf blight, caused by Cochliobolus heterostrophus race O, and flowering time using advanced intercross maize lines. Genetics 176: 645–657 Bao W, O’Malley DM, Whetten R, Sederoff RR (1993) A laccase associated with lignification in loblolly pine xylem. Science 260: 672–674 Beckman PM, Payne GA (1982) External growth penetration and development of Cercospora zeae-maydis in corn Zea mays leaves. Phytopathology 72: 810–815 Benson JM, Poland JA, Benson BM, Stromberg EL, Nelson RJ (2015) Resistance to gray leaf spot of maize: genetic architecture and mechanisms elucidated through nested association mapping and near-isogenic line analysis. PLoS Genet 11: e1005045 Bent AF, Mackey D (2007) Elicitors, effectors, and R genes: the new paradigm and a lifetime supply of questions. Annu Rev Phytopathol 45: 399–436 Bertorello AM, Zhu JK (2009) SIK1/SOS2 networks: decoding sodium signals via calcium-responsive protein kinase pathways. Pflugers Arch 458: 613–619 Bhattacharjee S (2012) The language of reactive oxygen species signaling in plants. J Bot 2012: 985298 Bhuiyan NH, Selvaraj G, Wei Y, King J (2009) Role of lignification in plant defense. Plant Signal Behav 4: 158–159 Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES (2007) TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 23: 2633 Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72: 248–254 Brown JKM (2002) Yield penalties of disease resistance in crops. Curr Opin Plant Biol 5: 339–344 Buckler ES, Holland JB, Bradbury PJ, Acharya CB, Brown PJ, Browne C, Ersoz E, Flint-Garcia S, Garcia A, Glaubitz JC, et al (2009) The genetic architecture of maize flowering time. Science 325: 714–718 Buckner B, Johal GS, Janick-Buckner D (2000) Cell death in maize. Physiol Plant 108: 231–239 Plant Physiol. Vol. 172, 2016

Chaikam V, Negeri A, Dhawan R, Puchaka B, Ji J, Chintamanani S, Gachomo EW, Zillmer A, Doran T, Weil C, et al (2011) Use of MutantAssisted Gene Identification and Characterization (MAGIC) to identify novel genetic loci that modify the maize hypersensitive response. Theor Appl Genet 123: 985–997 Chen Z, Zheng Z, Huang J, Lai Z, Fan B (2009) Biosynthesis of salicylic acid in plants. Plant Signal Behav 4: 493–496 Chintamanani S, Hulbert SH, Johal GS, Balint-Kurti PJ (2010) Identification of a maize locus that modulates the hypersensitive defense response, using mutant-assisted gene identification and characterization. Genetics 184: 813–825 Coll NS, Epple P, Dangl JL (2011) Programmed cell death in the plant immune system. Cell Death Differ 18: 1247–1256 Cook JP, McMullen MD, Holland JB, Tian F, Bradbury P, Ross-Ibarra J, Buckler ES, Flint-Garcia SA (2012) Genetic architecture of maize kernel composition in the nested association mapping and inbred association panels. Plant Physiol 158: 824–834 Creissen G, Firmin J, Fryer M, Kular B, Leyland N, Reynolds H, Pastori G, Wellburn F, Baker N, Wellburn A, et al (1999) Elevated glutathione biosynthetic capacity in the chloroplasts of transgenic tobacco plants paradoxically causes increased oxidative stress. Plant Cell 11: 1277–1292 Driouich A, Laine AC, Vian B, Faye L (1992) Characterization and localization of laccase forms in stem and cell-cultures of sycamore. Plant J 2: 13–24 Elhiti M, Stasolla C (2009) Structure and function of homodomain-leucine zipper (HD-Zip) proteins. Plant Signal Behav 4: 86–88 Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES, Mitchell SE (2011) A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species. PLoS ONE 6: e19379 Ferdinandsen C, Winge Ö (1930) A heritable blotch leaf in oats. Hereditas 13: 164–176 Flint-Garcia SA, Thuillet AC, Yu J, Pressoir G, Romero SM, Mitchell SE, Doebley J, Kresovich S, Goodman MM, Buckler ES (2005) Maize association population: a high-resolution platform for quantitative trait locus dissection. Plant J 44: 1054–1064 Frei dit Frey N, Robatzek S (2009) Trafficking vesicles: pro or contra pathogens? Curr Opin Plant Biol 12: 437–443 Fu D, Uauy C, Distelfeld A, Blechl A, Epstein L, Chen X, Sela H, Fahima T, Dubcovsky J (2009) A kinase-START gene confers temperaturedependent resistance to wheat stripe rust. Science 323: 1357–1360 Gallego-Giraldo L, Escamilla-Trevino L, Jackson LA, Dixon RA (2011) Salicylic acid mediates the reduced growth of lignin down-regulated plants. Proc Natl Acad Sci USA 108: 20814–20819 Ganal MW, Durstewitz G, Polley A, Bérard A, Buckler ES, Charcosset A, Clarke JD, Graner EM, Hansen M, Joets J, et al (2011) A large maize (Zea mays L.) SNP genotyping array: development and germplasm genotyping, and genetic mapping to compare with the B73 reference genome. PLoS ONE 6: e28334 Gechev TS, Gadjev IZ, Hille J (2004) An extensive microarray analysis of AAL-toxin-induced cell death in Arabidopsis thaliana brings new insights into the complexity of programmed cell death in plants. Cell Mol Life Sci 61: 1185–1197 Gerdes JT, Tracy WF (1993) Pedigree diversity within the Lancaster Surecrop Heterotic group of maize. Crop Sci 33: 334–337 Gray J, Close PS, Briggs SP, Johal GS (1997) A novel suppressor of cell death in plants encoded by the Lls1 gene of maize. Cell 89: 25–31 Gray J, Janick-Buckner D, Buckner B, Close PS, Johal GS (2002) Lightdependent death of maize lls1 cells is mediated by mature chloroplasts. Plant Physiol 130: 1894–1907 Ham BK, Park JM, Lee SB, Kim MJ, Lee IJ, Kim KJ, Kwon CS, Paek KH (2006) Tobacco Tsip1, a DnaJ-type Zn finger protein, is recruited to and potentiates Tsi1-mediated transcriptional activation. Plant Cell 18: 2005– 2020 Hammond-Kosack KE, Jones JD (1996) Resistance gene-dependent plant defense responses. Plant Cell 8: 1773–1791 Hirschhorn JN, Daly MJ (2005) Genome-wide association studies for common diseases and complex traits. Nat Rev Genet 6: 95–108 Holland JB (1998) EPISTACY: a SAS program for detecting two-locus epistatic interactions using genetic marker information. J Hered 89: 374–375 Hu G, Richter TE, Hulbert SH, Pryor T (1996) Disease lesion mimicry caused by mutations in the rust resistance gene Rp1. Plant Cell 8: 1367– 1376 1801

Olukolu et al.

Hückelhoven R (2007) Cell wall-associated mechanisms of disease resistance and susceptibility. Annu Rev Phytopathol 45: 101–127 Hung CY, Aspesi P Jr, Hunter MR, Lomax AW, Perera IY (2014) Phosphoinositide-signaling is one component of a robust plant defense response. Front Plant Sci 5: 267 Hung HY, Browne C, Guill K, Coles N, Eller M, Garcia A, Lepak N, Melia-Hancock S, Oropeza-Rosas M, Salvo S, et al (2012) The relationship between parental genetic or phenotypic divergence and progeny variation in the maize nested association mapping population. Heredity 108: 490–499 Jambunathan N (2010) Determination and detection of reactive oxygen species (ROS), lipid peroxidation, and electrolyte leakage in plants. Methods Mol Biol 639: 291–297 Jennings P, Ullstrup AJ (1957) A histological study of three Helminthosporium leaf blights on corn. Phytopathology 47: 707–714 Johal GS(2007) Disease lesion mimic mutants of maize. American Phytopathological Society Feature Story. www.apsnet.org/publications/ apsnetfeatures/Pages/MutantsofMaize.aspx Johal GS, Hulbert SH, Briggs SP (1995) Disease lesion mimics of maize: a model for cell-death in plants. BioEssays 17: 685–692 Kang HM, Zaitlen NA, Wade CM, Kirby A, Heckerman D, Daly MJ, Eskin E (2008) Efficient control of population structure in model organism association mapping. Genetics 178: 1709–1723 Knox-Davies PS (1974) Penetration of maize leaves by Helminthosporium turcicum. Phytopathology 64: 1468–1470 Kump KL, Bradbury PJ, Wisser RJ, Buckler ES, Belcher AR, OropezaRosas MA, Zwonitzer JC, Kresovich S, McMullen MD, Ware D, et al (2011) Genome-wide association study of quantitative resistance to southern leaf blight in the maize nested association mapping population. Nat Genet 43: 163–168 Larsson SJ, Lipka AE, Buckler ES (2013) Lessons from Dwarf8 on the strengths and weaknesses of structured association mapping. PLoS Genet 9: e1003246 Li D, Zhang H, Song Q, Wang L, Liu S, Hong Y, Huang L, Song F (2015) Tomato Sl3-MMP, a member of the matrix metalloproteinase family, is required for disease resistance against Botrytis cinerea and Pseudomonas syringae pv. tomato DC3000. BMC Plant Biol 15: 143 Li E, Wang S, Liu Y, Chen JG, Douglas CJ (2011) OVATE FAMILY PROTEIN4 (OFP4) interaction with KNAT7 regulates secondary cell wall formation in Arabidopsis thaliana. Plant J 67: 328–341 Lorrain S, Vailleau F, Balagué C, Roby D (2003) Lesion mimic mutants: keys for deciphering cell death and defense pathways in plants? Trends Plant Sci 8: 263–271 Mach JM, Castillo AR, Hoogstraten R, Greenberg JT (2001) The Arabidopsis-accelerated cell death gene ACD2 encodes red chlorophyll catabolite reductase and suppresses the spread of disease symptoms. Proc Natl Acad Sci USA 98: 771–776 MacMillan CP, Mansfield SD, Stachurski ZH, Evans R, Southerton SG (2010) Fasciclin-like arabinogalactan proteins: specialization for stem biomechanics and cell wall architecture in Arabidopsis and Eucalyptus. Plant J 62: 689–703 MacMillan CP, Taylor L, Bi Y, Southerton SG, Evans R, Spokevicius A (2015) The fasciclin-like arabinogalactan protein family of Eucalyptus grandis contains members that impact wood biology and biomechanics. New Phytol 206: 1314–1327 Makepeace JC, Oxley SJP, Havis ND, Hackett R, Burke JI, Brown JKM (2007) Associations between fungal and abiotic leaf spotting and the presence of mlo alleles in barley. Plant Pathol 56: 934–942 Maleck K, Neuenschwander U, Cade RM, Dietrich RA, Dangl JL, Ryals JA (2002) Isolation and characterization of broad-spectrum diseaseresistant Arabidopsis mutants. Genetics 160: 1661–1671 Malinovsky FG, Fangel JU, Willats WGT (2014) The role of the cell wall in plant immunity. Front Plant Sci 5: 178 Marchler-Bauer A, Anderson JB, Cherukuri PF, DeWeese-Scott C, Geer LY, Gwadz M, He S, Hurwitz DI, Jackson JD, Ke Z, et al (2005) CDD: a conserved domain database for protein classification. Nucleic Acids Res 33: D192–D196 Marshansky V, Futai M (2008) The V-type H+-ATPase in vesicular trafficking: targeting, regulation and function. Curr Opin Cell Biol 20: 415–426 Maschietto V, Marocco A, Malachova A, Lanubile A (2015) Resistance to Fusarium verticillioides and fumonisin accumulation in maize inbred lines involves an earlier and enhanced expression of lipoxygenase (LOX) genes. J Plant Physiol 188: 9–18 1802

McMullen MD, Kresovich S, Villeda HS, Bradbury P, Li H, Sun Q, FlintGarcia S, Thornsberry J, Acharya C, Bottoms C, et al (2009) Genetic properties of the maize nested association mapping population. Science 325: 737–740 Meskiene I, Baudouin E, Schweighofer A, Liwosz A, Jonak C, Rodriguez PL, Jelinek H, Hirt H (2003) Stress-induced protein phosphatase 2C is a negative regulator of a mitogen-activated protein kinase. J Biol Chem 278: 18945–18952 Mittler R (2002) Oxidative stress, antioxidants and stress tolerance. Trends Plant Sci 7: 405–410 Morley SJ, Coldwell MJ, Clemens MJ (2005) Initiation factor modifications in the preapoptotic phase. Cell Death Differ 12: 571–584 Nair S, Tomar SMS (2001) Genetical and anatomical analyses of a leaf flecking mutant in Triticum aestivum L. Euphytica 121: 53–58 Negeri A, Wang GF, Benavente L, Kibiti CM, Chaikam V, Johal G, BalintKurti P (2013) Characterization of temperature and light effects on the defense response phenotypes associated with the maize Rp1-D21 autoactive resistance gene. BMC Plant Biol 13: 106 Neuffer MG, Calvert OH (1975) Dominant disease lesion mimics in maize. J Hered 66: 265–270 Noctor G, De Paepe R, Foyer CH (2007) Mitochondrial redox biology and homeostasis in plants. Trends Plant Sci 12: 125–134 Olukolu BA, Negeri A, Dhawan R, Venkata BP, Sharma P, Garg A, Gachomo E, Marla S, Chu K, Hasan A, et al (2013) A connected set of genes associated with programmed cell death implicated in controlling the hypersensitive response in maize. Genetics 193: 609–620 Olukolu BA, Tracy WF, Wisser R, De Vries B, Balint-Kurti PJ (2016) A genome-wide association study for partial resistance to maize common rust. Phytopathology 106: 745–751 Olukolu BA, Wang GF, Vontimitta V, Venkata BP, Marla S, Ji J, Gachomo E, Chu K, Negeri A, Benson J, et al (2014) A genome-wide association study of the maize hypersensitive defense response identifies genes that cluster in related pathways. PLoS Genet 10: e1004562 Panagiotou OA, Ioannidis JP (2012) What should the genome-wide significance threshold be? Empirical replication of borderline genetic associations. Int J Epidemiol 41: 273–286 Papp E, Nardai G, Söti C, Csermely P (2003) Molecular chaperones, stress proteins and redox homeostasis. Biofactors 17: 249–257 Penning BW, Johal GS, McMullen MD (2004) A major suppressor of cell death, slm1, modifies the expression of the maize (Zea mays L.) lesion mimic mutation les23. Genome 47: 961–969 Piepho HP, Williams ER, Fleck M (2006) A note on the analysis of designed experiments with complex treatment structure. HortScience 41: 446–452 Poland JA, Bradbury PJ, Buckler ES, Nelson RJ (2011) Genome-wide nested association mapping of quantitative resistance to northern leaf blight in maize. Proc Natl Acad Sci USA 108: 6893–6898 Romay MC, Millard MJ, Glaubitz JC, Peiffer JA, Swarts KL, Casstevens TM, Elshire RJ, Acharya CB, Mitchell SE, Flint-Garcia SA, et al (2013) Comprehensive genotyping of the USA national maize inbred seed bank. Genome Biol 14: R55 Rustérucci C, Aviv DH, Holt BF III, Dangl JL, Parker JE (2001) The disease resistance signaling components EDS1 and PAD4 are essential regulators of the cell death pathway controlled by LSD1 in Arabidopsis. Plant Cell 13: 2211–2224 Samayoa LF, Malvar RA, Olukolu BA, Holland JB, Butrón A (2015) Genome-wide association study reveals a set of genes associated with resistance to the Mediterranean corn borer (Sesamia nonagrioides L.) in a maize diversity panel. BMC Plant Biol 15: 35 SAS Institute I (2000-2004) SAS 9.2 Help and Documentation. SAS, Cary, NC Schnable PS, Ware D, Fulton RS, Stein JC, Wei F, Pasternak S, Liang C, Zhang J, Fulton L, Graves TA, et al (2009) The B73 maize genome: complexity, diversity, and dynamics. Science 326: 1112–1115 Schuler MA (1996) Plant cytochrome P450 monooxygenases. Crit Rev Plant Sci 15: 235–284 Schweighofer A, Kazanaviciute V, Scheikl E, Teige M, Doczi R, Hirt H, Schwanninger M, Kant M, Schuurink R, Mauch F, et al (2007) The PP2C-type phosphatase AP2C1, which negatively regulates MPK4 and MPK6, modulates innate immunity, jasmonic acid, and ethylene levels in Arabidopsis. Plant Cell 19: 2213–2224 Shah J (2005) Lipids, lipases, and lipid-modifying enzymes in plant disease resistance. Annu Rev Phytopathol 43: 229–260 Plant Physiol. Vol. 172, 2016

Genetic Control of Maize Leaf Flecking

Shaner G, Finney PE (1977) The effect of nitrogen fertilizer on expression of slow mildewing resistance in Knox wheat. Phytopathology 67: 1051– 1056 Silva LdaC, Wang S, Zeng ZB (2012) Composite interval mapping and multiple interval mapping: procedures and guidelines for using Windows QTL Cartographer. 871: 75–119 Sterjiades R, Dean JF, Eriksson KE (1992) Laccase from sycamore maple (Acer pseudoplatanus) polymerizes monolignols. Plant Physiol 99: 1162– 1168 Storey JD, Tibshirani R (2003) Statistical significance for genomewide studies. Proc Natl Acad Sci USA 100: 9440–9445 Tian F, Bradbury PJ, Brown PJ, Hung H, Sun Q, Flint-Garcia S, Rocheford TR, McMullen MD, Holland JB, Buckler ES (2011) Genome-wide association study of leaf architecture in the maize nested association mapping population. Nat Genet 43: 159–162 Vainberg IE, Lewis SA, Rommelaere H, Ampe C, Vandekerckhove J, Klein HL, Cowan NJ (1998) Prefoldin, a chaperone that delivers unfolded proteins to cytosolic chaperonin. Cell 93: 863–873 Valdar W, Holmes CC, Mott R, Flint J (2009) Mapping in structured populations by resample model averaging. Genetics 182: 1263–1277 van Doorn WG, Beers EP, Dangl JL, Franklin-Tong VE, Gallois P, HaraNishimura I, Jones AM, Kawai-Yamada M, Lam E, Mundy J, et al (2011) Morphological classification of plant cell deaths. Cell Death Differ 18: 1241–1246 Vontimitta V, Olukolu BA, Penning BW, Johal G, Balint-Kurti PJ (2015) The genetic basis of flecking and its relationship to disease resistance in the IBM maize mapping population. Theor Appl Genet 128: 2331–2339 Voorrips RE (2002) MapChart: software for the graphical presentation of linkage maps and QTLs. J Hered 93: 77–78 Walbot V, Hoisington D, Neuffer MG (1983) Disease lesion mimic mutations. In T Kosuge, CP Meredith, A Hollaender, eds, Genetic Engineering of Plants, Vol 26. Plenum Press, New York, pp 431–432 Wallace JG, Bradbury PJ, Zhang N, Gibon Y, Stitt M, Buckler ES (2014) Association mapping across numerous traits reveals patterns of functional variation in maize. PLoS Genet 10: e1004845 Wang GF, Balint-Kurti PJ (2016) Maize homologs of CCoAOMT and HCT, two key enzymes in lignin biosynthesis, form complexes with the NLR Rp1 protein to modulate the defense response. Plant Physiol 171: 2166–2177

Plant Physiol. Vol. 172, 2016

Wang S, Chang Y, Guo J, Zeng Q, Ellis BE, Chen JG (2011) Arabidopsis ovate family proteins, a novel transcriptional repressor family, control multiple aspects of plant growth and development. PLoS ONE 6: e23896 Wisser RJ, Kolkman JM, Patzoldt ME, Holland JB, Yu J, Krakowsky M, Nelson RJ, Balint-Kurti PJ (2011) Multivariate analysis of maize disease resistances suggests a pleiotropic genetic basis and implicates a GST gene. Proc Natl Acad Sci USA 108: 7339–7344 Wolter M, Hollricher K, Salamini F, Schulze-Lefert P (1993) The mlo resistance alleles to powdery mildew infection in barley trigger a developmentally controlled defence mimic phenotype. Mol Gen Genet 239: 122–128 Wu J, Baldwin IT (2010) New insights into plant responses to the attack from insect herbivores. Annu Rev Genet 44: 1–24 Yanagisawa S, Sheen J (1998) Involvement of maize Dof zinc finger proteins in tissue-specific and light-regulated gene expression. Plant Cell 10: 75–89 Yin Z, Chen J, Zeng L, Goh M, Leung H, Khush GS, Wang GL (2000) Characterizing rice lesion mimic mutants and identifying a mutant with broad-spectrum resistance to rice blast and bacterial blight. Mol Plant Microbe Interact 13: 869–876 Yu J, Pressoir G, Briggs WH, Vroh Bi I, Yamasaki M, Doebley JF, McMullen MD, Gaut BS, Nielsen DM, Holland JB, et al (2006) A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nat Genet 38: 203–208 Zehr BE, Dudley JW, Rufener GK (1994) QTLs for degree of pollen-silk discordance, expression of disease lesion mimic, and leaf curl response to drought. Maize Genet Coop Newsl 68: 109–200 Zhang Z, Ersoz E, Lai CQ, Todhunter RJ, Tiwari HK, Gore MA, Bradbury PJ, Yu J, Arnett DK, Ordovas JM, et al (2010) Mixed linear model approach adapted for genome-wide association studies. Nat Genet 42: 355–360 Zhang Z, Liu Y, Ding P, Li Y, Kong Q, Zhang Y (2014) Splicing of receptor-like kinase-encoding SNC4 and CERK1 is regulated by two conserved splicing factors that are required for plant immunity. Mol Plant 7: 1766–1775 Zila CT, Ogut F, Romay MC, Gardner CA, Buckler ES, Holland JB (2014) Genome-wide association study of Fusarium ear rot disease in the U.S. A. maize inbred line collection. BMC Plant Biol 14: 372 Zila CT, Samayoa LF, Santiago R, Butron A, Holland JB (2013) A genomewide association study reveals genes associated with Fusarium ear rot resistance in a maize core diversity panel. G3 (Bethesda) 3: 2095–2104

1803