The Second Site Modifier, Sympathy for the ligule, Encodes a Homolog of Arabidopsis ENHANCED DISEASE RESISTANCE4 and Rescues the Liguleless narrow Maize Mutant

Allelic variation at Sympathy for the ligule affects growth and development of Liguleless narrow maize mutants.

Abstract

Liguleless narrow1 encodes a plasma membrane-localized receptor-like kinase required for normal development of maize (Zea mays) leaves, internodes, and inflorescences. The semidominant Lgn-R mutation lacks kinase activity, and phenotypic severity is dependent on inbred background. We created near isogenic lines and assayed the phenotype in multiple environments. Lgn-R plants that carry the B73 version of Sympathy for the ligule (Sol-B) fail to grow under hot conditions, but those that carry the Mo17 version (Sol-M) survive at hot temperatures and are significantly taller at cool temperatures. To identify Sol, we used recombinant mapping and analyzed the Lgn-R phenotype in additional inbred backgrounds. We identified amino acid sequence variations in GRMZM2G075262 that segregate with severity of the Lgn-R phenotypes. This gene is expressed at high levels in Lgn-R B73, but expression drops to nonmutant levels with one copy of Sol-M. An EMS mutation solidified the identity of SOL as a maize homolog of Arabidopsis (Arabidopsis thaliana) ENHANCED DISEASE RESISTANCE4 (EDR4). SOL, like EDR4, is induced in response to pathogen-associated molecular patterns such as flg22. Integrated transcriptomic and phosphoproteomic analyses suggest that Lgn-R plants constitutively activate an immune signaling cascade that induces temperature-sensitive responses in addition to defects in leaf development. We propose that aspects of the severe Lgn-R developmental phenotype result from constitutive defense induction and that SOL potentially functions in repressing this response in Mo17 but not B73. Identification of LGN and its interaction with SOL provides insight into the integration of developmental control and immune responses.

INTRODUCTION

The expressivity of a mutant phenotype is often dependent on other genes. These second site modifiers have been identified through mutagenesis screens and crosses to different backgrounds. Background-dependent modifiers have been found for developmental pathways as diverse as tomato (Solanum lycopersicum) fruit size and Drosophila (Drosophila melanogaster) bristle number (Gibert et al., 2005; Rodríguez et al., 2013). Maize (Zea mays) is particularly rich for the identification of modifiers because of the high genetic variation among maize inbred lines: to illustrate this point, the coding sequence variation between two maize inbreds is as great as the coding sequence variation between humans and chimpanzees (Pan troglodytes; Buckler et al., 2006), and the noncoding intergenic space is also remarkably variable between inbreds (Brunner et al., 2005). Examples of modifiers in maize include those for seed protein content (Babu et al., 2015), lesion mimic expressivity (Penning et al., 2004), and virescence (Xing et al., 2014). Few modifiers that affect pleiotropic phenotypes, however, have been analyzed at the molecular level.

Liguleless narrow-R (R for reference allele) has striking developmental phenotypes caused by an EMS-induced point mutation that eliminates protein kinase activity (Moon et al., 2013). As a heterozygote in the inbred line B73, Lgn-R plants are short with narrow leaves and reduced inflorescences. The adult leaves fail to properly develop ligules and auricles, structures that are located at the blade/sheath boundary in grasses. The ligule keeps water and debris from entering into the stem where the axillary bud forms (Chaffey, 2000). The auricle allows the blade to tilt back and maximize exposure to sunlight.

The expressivity of the Lgn-R phenotype is background-dependent. While the phenotype is severe in B73, plant height, leaf width, and inflorescence and ligule development are all restored to near wild-type levels in Mo17 (Buescher et al., 2014). To investigate these background differences, we crossed Lgn-R in B73 to the intermated Mo17 × B73 recombinant inbred lines (IBM lines; Lee et al., 2002) and conducted a quantitative trait locus (QTL) analysis. Sympathy for the ligule (Sol) was identified as a main effect QTL on chromosome 1 (Buescher et al., 2014). A genotype-by-environment (GxE) effect was discovered by growing the population in two different environments. Lgn-R B73 mutants failed to grow in the hot summers of West Lafayette, Indiana, but survived to reproduce in the cooler weather of Albany, California. Phenotypic expressivity that is dependent on background and temperature is often seen in autoimmune mutants in other plant species (van Wersch et al., 2016).

Here, we describe Lgn-R in the near isogenic line (NIL) that contains Mo17 at Sol in a predominantly B73 background and identify the gene for Sol by fine-mapping and expression analysis. Sol encodes a homolog of Arabidopsis (Arabidopsis thaliana) ENHANCED DISEASE RESISTANCE4 (EDR4). EDR4 binds to EDR1, a MAP kinase kinase kinase (MAP3K), and localizes it to the site of hyphal penetration pegs (Wu et al., 2015). Like EDR4, Sol is transcriptionally expressed in response to treatment with pathogen-associated molecule patterns (PAMPs). Sol transcript levels are also increased in expression under normal growth conditions in Lgn-R B73 but not in Lgn-R NIL and nonmutant B73 or Mo17 siblings. Thus, the NIL has both developmental phenotypes and Sol expression levels that more closely resemble its nonmutant siblings. Based on transcriptomic, proteomic, and phenotypic analyses, we hypothesize that the Lgn-R mutation triggers an autoimmune syndrome and that the NIL modifies this response. Our data begin to highlight hypotheses to explain the modifying behavior of Sol on Lgn-R and identify important molecular components that contribute to the Lgn-R phenotype.

RESULTS

The NIL Dampens the Lgn-R Mutant Phenotype

To investigate the difference in expressivity between B73 and Mo17, we created NILs by first crossing Lgn-R to the IBM recombinant inbreds (Buescher et al., 2014) and then backcrossing the least severe Lgn-R individuals to B73 for a minimum of four generations. We refer to the Mo17 Sol locus as Sol-M and the B73 locus as Sol-B. Phenotypes were assessed in Lgn-R heterozygotes, and plants were either Sol-M/Sol-B (NIL) or Sol-B/Sol-B (B73).

We grew Lgn-R NIL plants over multiple seasons to determine the effect on plant architecture in comparison with Lgn-R in the B73 and Mo17 backgrounds (Figures 1A to 1C). Whereas Lgn-R Mo17 has nearly the same height and leaf width as the nonmutant Mo17 inbred, both height and width are reduced in the NIL, and most severe in B73 (Figures 1D to 1G; Supplemental Figures 1A and 1B). In Albany, California, nonmutant plants averaged 179.4 ± 10.5 cm in height and 10 ± 0.6 cm in leaf width, Lgn-R NIL plants averaged 147.4 ± 10.2 cm in height and 7.5 ± 0.9 cm in leaf width, and Lgn-R B73 averaged only 102.4 ± 22.4 cm for plant height and 4.1 ± 0.7 cm for leaf width. Ear development was also restored in Lgn-R NIL plants compared with Lgn-R B73 (Figures 1A and 1H).

Figure 1.

The Sol-M Modifier Rescues the Lgn-R Phenotype.

(A) Mature plant phenotypes pictured from left to right: Mo17, B73, Lgn-R Mo17, Lgn-R NIL, and Lgn-R B73. Ears are circled in red.

(B) Adaxial view of ligules from leaf 6. The Lgn-R B73 ligule is in the red box.

(C) Abaxial view of ligules from leaf 6 in same order as in (B). Auricles are marked by red angles.

(D) to (G) Plant height and leaf width measurements. Plants that died are represented on graphs with 0 cm height and 0 cm width. The number of dead is shown with n values.

(D) Plant height measurements for Lgn-R segregating 1:1 in Mo17 and B73. A one-way ANOVA with a Tukey’s posthoc test (95% CI) showed that Lgn-R B73 is statistically more severe than Lgn-R Mo17 in Albany (P < 1e-7). In Davis, the proportion of dead plants per genotype was found to be statistically significant by z test for Lgn-R B73 plants compared with Lgn-R Mo17 (z = −35.0, P < 0.00001).

(E) Plant leaf width measurements for Lgn-R segregating 1:1 in Mo17 and B73. A one-way ANOVA with a Tukey’s posthoc test (95% CI) showed that Lgn-R B73 is statistically more severe than Lgn-R Mo17 in Albany (P < 1e-7).

(F) Plant height measurements for Lgn-R segregating 1:1 in NIL (Sol-B/Sol-M) and B73 (Sol-B/Sol-B) backgrounds. A one-way ANOVA with a Tukey’s posthoc test (95% CI) showed significant differences between Lgn-R NIL and Lgn-R B73 plants (P = 0.02). In Davis, the proportion of dead plants per genotype was found to be statistically significant by z test for Lgn-R NIL plants compared with Lgn-R B73 (z = −8.25, P < 0.00001).

(G) Plant leaf width measurements for Lgn-R segregating 1:1 in NIL and B73 backgrounds. A one-way ANOVA with a Tukey’s posthoc test (95% CI) showed significant differences between Lgn-R NIL and Lgn-R B73 plants (P = 3.5e-6).

(H) Percentage of plants that develop ears. A z test found that the proportion of Lgn-R NIL plants with ears is significantly greater than the proportion of Lgn-R B73 plants that form ears (z = −31.0, P < 0.00001). Error bars represent se.

(I) and (J) Plant height and leaf measurements for a family segregating for Lgn-R and Sol. A one-way ANOVA with a Tukey’s posthoc test (95% CI) found significant differences between heterozygous and homozygous Lgn-R regardless of Sol genotype.

(I) Within Lgn-R heterozygotes, the NIL and Sol-M/Sol-M plants have greater plant heights than their B73 siblings (P < 0.007 and P < 0.004), but there are not significant differences between the NIL and Sol-M/Sol-M individuals. Among the Lgn-R homozygotes, the NIL has statistically shorter heights than Sol-M/Sol-M plants (P < 0.02) and is significantly taller than Sol-B/Sol-B (P < 0.007).

(J) Within Lgn-R heterozygotes, the NIL and Sol-M/Sol-M plants have greater leaf widths than Sol-B/Sol-B (P < 2e-4 and P < 2e-6) but are not significantly different from each other. Within Lgn-R homozygotes, we see a clear dosage effect with each copy of Sol-M, with heterozygotes significantly wider than Sol-B/Sol-B plants (P < 0.004) and significantly narrower than Sol-M/Sol-M plants (P < 9e-4).

(K) Adaxial view of the ligule region of Lgn-R heterozygotes in order from top to bottom of Sol-M/Sol-M, Sol-M/Sol-B, and Sol-B/Sol-B.

Nonmutant B73 and Mo17 plants as well as Lgn-R Mo17 mutants had morphologically normal ligules and auricles located between distinct blade and sheath regions. The ligule in Lgn-R NIL plants formed across the leaf but the auricle was only partially restored and, as previously described (Moon et al., 2013), Lgn-R B73 plants never developed a complete ligule or auricle in adult leaves (Figures 1B and 1C).

To elucidate the dosage effect of Sol-M on Lgn-R in the B73 background, we self-pollinated the Lgn-R NIL and analyzed the segregating genotypes in the greenhouse. Of the Lgn-R heterozygotes, the NIL and Sol-M/Sol-M plants had greater plant heights and leaf widths than their B73 siblings (height and width: P < 2e-4 and P < 2e-6), but there were no significant differences between the NIL and Sol-M/Sol-M individuals (Figures 1I and 1J; Supplemental Figure 1C). The ligule itself, however, was near normal, with two copies of Sol-M (Figure 1K). Among the Lgn-R homozygotes, the Sol-M allele exhibited a clear dosage effect in that the NIL had statistically greater heights and widths than Sol-B/Sol-B individuals (P < 0.007 and P < 0.004) but had significantly smaller heights and widths than Sol-M/Sol-M plants (P < 0.02 and P < 0e-00; Figures 1I and 1J; Supplemental Figure 1C). Thus, one copy of Sol-M can improve the growth of both Lgn-R heterozygotes and homozygotes, but Lgn-R homozygotes retain multiple defects.

Sol Rescues Temperature-Dependent Lgn-R Lethality

A GxE effect was observed when the original Lgn-R × IBM lines were analyzed in West Lafayette, Indiana, and Albany, California (Buescher et al., 2014). To explore the underlying cause, plants were grown at different temperatures. Lgn-R NIL plants segregating for Sol-M were grown in Davis, California, where average daytime temperatures during our field seasons were 33.1°C compared with that of West Lafayette, Indiana, at 29.0°C and Albany, California, at 23.2°C (data available at http://www.ncdc.noaa.gov). Nights have similar temperatures in Davis and Albany and are slightly warmer in West Lafayette.

Lgn-R plants showed significant GxE effects between the Albany, California, and Davis, California, fields (Figures 1D to 1G). A one-way ANOVA with a Tukey’s posthoc test (95% confidence interval [CI]) found significant effects from genotype (P = 2e-16), location (P = 2e-16), and genotype-by-location interaction (P = 1.4e-11) when comparing the Lgn-R B73 and Lgn-R Mo17 populations as well as significant effects from genotype (P = 2.7e-11), location (P = 0.0019), and genotype-by-location interaction (P = 4e-7) when comparing Lgn-R NIL and Lgn-R B73 siblings. All pairwise comparisons can be found in Supplemental Figures 1A and 1B. For example, during the 2015 Davis field season, 17 out of 19 Lgn-R plants that were homozygous B73 at the Sol locus died within 5 weeks postgermination (Figures 2A and 2B). By contrast, only 1 out of 10 Lgn-R NIL plants died. None of the Lgn-R plants died in Albany. The Lgn-R NIL plants that survived in Davis were shorter and had narrower leaves compared with their Albany counterparts. Nonmutant plants survived at both locations and were, in fact, taller in Davis (Figure 2B).

Figure 2.

Lgn-R Plants Survive the Heat with the Sol-M Modifier.

(A) Mature plant phenotypes of individuals segregating for Lgn-R and Sol-M in Davis. Dead plants (circled in white) are Lgn-R B73 and flank an Lgn-R NIL plant.

(B) Height measurements of individuals segregating for Lgn-R and Sol-M in cool (Albany) versus hot (Davis) climate. Nonmutant siblings include both Sol genotypes. A two-tailed t test showed that Lgn-R NIL plants were significantly different between the two fields [t(49) = 9.1, P = 4e-12], as were Lgn-R B73 plants [t(47) = 15.2, P = 9e-20].

(C) Top, from left to right: wild-type, Lgn-R NIL, and Lgn-R B73 plants at 30 d after planting in the cycling cool growth chamber (11–21°C). At bottom is same order in the cycling hot growth chamber (15–32°C).

(D) Mature plants 30 d after cycling hot growth chamber treatment from left to right: nonmutant, Lgn-R NIL, and Lgn-R B73. Only Lgn-R B73 plants were unable to recover from heat treatment.

(E) Adaxial view of the ligule of leaf 6 when plants are grown at a constant 17°C.

(F) Adaxial view of the ligule of leaf 6 when plants are grown at a constant 30°C.

(G) Plant heights and leaf widths for a family grown at a constant 17°C compared with 30°C. A one-way ANOVA revealed significant differences in height between the nonmutant and Lgn-R B73 siblings at 30°C (P < 0.01) but not at 17°C, whereas significant differences in leaf width were seen at both temperatures when comparing Lgn-R in B73 with the NIL (P < 1e-4 and P < 5e-3) and nonmutant siblings.

(H) The change in plant height and leaf width observed per genotype in 17°C conditions compared with 30°C conditions. Two-tailed t tests showed that Lgn-R B73 plants were most significantly altered between the two temperature regimes compared with both nonmutant and NIL siblings in terms of both height [t(25) = 4.4, P < 2e-4 and t(13) = 2.5, P < 0.03] and width [t(25) = 2.3, P < 0.03 and t(13) = 3.1, P < 8e-3]. Lgn-R NIL lines were not significantly different from their nonmutant siblings.

Error bars in (B) and (H) represent se.

To determine whether growth at high temperature is sufficient to trigger Lgn-R lethality, we grew segregating Lgn-R NIL families under two different temperature regimes in otherwise constant growth chambers (15–32°C and 11–21°C). The cooler temperature within these ranges was maintained at night and the warmer temperature was used during the day in an attempt to mimic field conditions. In the cool cycling growth chamber, the genotypes had nearly identical phenotypes, whereas Lgn-R NIL plants were noticeably more severe in the warm cycling chamber and Lgn-R B73 plants were the most strongly affected (Figure 2C). After removal from the warm cycling chamber, nonmutant and Lgn-R NIL plants continued to grow and form reproductive tissues while Lgn-R B73 plants never recovered (Figure 2D). We also grew the segregating populations at either a constant 17 or 30°C. Significant reduction in plant height and leaf width was seen across all genotypes at 30°C compared with 17°C, with Lgn-R B73 the most strongly affected by the 30°C condition. Intriguingly, some of the phenotypic differences seen under normal field conditions were eliminated in the 17°C growth chamber experiment. Although leaves are still narrower in Lgn-R B73 compared with NIL and nonmutant siblings, plant height was not significantly different between any of the genotypes at this temperature and ligule development was restored in Lgn-R B73 plants at 17°C (Figures 2E to 2H; Supplemental Figure 1D). In summary, Lgn-R phenotypes are more severe at higher temperatures and rescued at cooler temperatures and in the presence of the Mo17 allele of Sol.

Sol Is an Ortholog of EDR4

To identify the Sol modifier, we genotyped plants from Lgn-R NIL families and scored the phenotypes (Supplemental Table 1; see “Materials and Methods”). With recombinant mapping techniques, we were able to place Sol between IDP1489 at 56,572K bp and bnlg2238 at 55,080K bp on chromosome 1 (Figure 3A). Based on the idea that syntenic genes are most likely to be expressed as proteins (Walley et al., 2016), we used the Comparative Genomics website to identify four maize genes in the interval that have orthologs in sorghum (Sorghum bicolor), rice (Oryza sativa), Setaria italica, and Brachypodium distachyon (Lyons and Freeling, 2008). The syntenic genes included GRMZM2G075262, a gene of unknown function, GRMZM2G072892, a putative LUNG-7 transmembrane receptor, GRMZM2G049211, a putative vacuolar sorting receptor precursor, and GRMZM2G119850, a putative receptor-like kinase.

Figure 3.

Map Position of Sol and Analysis of the Locus in NAM Founder Lines.

(A) Maize chromosome 1 with markers used for fine-mapping of the Sol QTL.

(B) Plant height of maize inbreds crossed once to Lgn-R/+ in B73 and grown in Indiana. Error bars represent se.

(C) to (I) Plant height and leaf width comparisons for families segregating Lgn-R and Sol in our cool (Albany) and hot (Davis) environments. Dead plants are represented on graphs with 0 cm height and 0 cm width, and number per genotype (n) is shown.

(C) Family segregating at Sol for Mo17 and Ms71 alleles; heterozygotes are significantly rescued compared with homozygotes. A one-way ANOVA with a Tukey’s posthoc test (95% CI) found a significant difference in leaf width in Albany (P = 1.3e-4). A z test performed on the proportion of dead in Davis also found significant differences (z = −11.8, P < 0.00001).

(D) Family segregating at Sol for B73 and CML228 alleles; heterozygotes are not statistically different from homozygotes at either location.

(E) Family segregating at Sol for B73 and CML247; heterozygotes are not statistically different from homozygotes at either location.

(F) Family segregating at Sol for B73 and Nc358; heterozygotes are not statistically different from homozygotes at either location.

(G) Family segregating at Sol for B73 and Nc350 alleles; heterozygotes display a significant rescued phenotype at both locations compared with homozygotes. A one-way ANOVA with a Tukey’s posthoc test (95% CI) found a significant difference in leaf width in Albany (P = 0.01). A z test performed on the proportion of dead in Davis also found significant differences (z = −27.5, P < 0.00001).

(H) Family segregating at Sol for B73 and Tzi8 alleles; heterozygotes display a significant rescued phenotype at both locations compared with homozygotes. A one-way ANOVA with a Tukey’s posthoc test (95% CI) found a significant difference in leaf width (P = 1.5e-5) and plant height (P = 7.3e-4) in Albany. A z test performed on the proportion of dead in Davis also found significant differences (z = −11.8, P < 0.00001).

(I) Family segregating at Sol for B73 and M162W. A one-way ANOVA with a Tukey’s posthoc test (95% CI) found a significant difference in leaf width (P = 0.01) in Albany. A z test performed on the proportion of dead heterozygotes compared with homozygotes in Davis also found significant differences (z = −14.9, P < 0.00001).

To continue investigating this interval, we crossed Lgn-R in B73 to the NAM Founder lines (McMullen et al., 2009) to determine which additional lines rescued the Lgn-R mutation at the Sol locus. Twenty F1 Lgn-R hybrid populations were grown in Indiana, where Lgn-R in B73 is severe. One Lgn-R hybrid, Ms71/B73, showed a lethal phenotype. All other hybrids rescued the Lgn-R phenotype to some extent (Figure 3B), although the rescued phenotype is not necessarily due to the Sol locus. Eight of the hybrids were successively backcrossed a minimum of three generations to B73 using the most rescued Lgn-R plant in each generation. Given the severe phenotype of Lgn-R in the Ms71/B73 hybrid, we also crossed Lgn-R in the Mo17 background to Ms71 and backcrossed to Ms71 at least two generations prior to analysis. To assess environment and background effects, we planted these segregating families in Davis and Albany during three field seasons. We determined plant height, leaf width, and the genotype of Sol for each family. We determined that a line rescued Lgn-R if it displayed increased plant height and leaf width when heterozygous at the Sol locus compared with Sol-B/Sol-B or Sol-Ms71/Sol-Ms71 siblings.

Of the NAM founder lines examined, four did not rescue Lgn-R at Sol (Ms71, CML228, CML247, and Nc358) and three lines displayed rescued phenotypes (Nc350, Tzi8, and M162W; Figures 3C to 3I; Supplemental Figures 1E to 1K). Of the severe lines, Lgn-R in the Ms71 inbred behaved similarly to Lgn-R B73. The average plant height and leaf width measurements for Lgn-R Sol-Ms71/Sol-Ms71 in Albany were 144.2 ± 35.7 and 5.5 ± 1.1 cm and in Davis they averaged 40.7 ± 43.5 and 3.2 ± 1.8 cm compared with the Lgn-R Sol-M/Sol-Ms71 measurements of 156 ± 30.6 and 8.9 ± 0.7 cm in Albany and 112.4 ± 26.4 and 5.1 ± 1.3 cm in Davis (Figure 3C). A one-way ANOVA followed by a Tukey’s posthoc test (95% CI) found significant effects from genotype (P = 8.8e-7), location (P = 0.0004), and genotype-by-location interaction (P = 0.007) for the family. The significance for individual comparisons can be found in Supplemental Figure 1E. CML228 was also not capable of rescuing Lgn-R at Sol. Sol-CML228/Sol-B plants have heights and widths that are not statistically different from Sol-B/Sol-B relatives at either field, although significant GxE effects were still observed for individuals within a given genotype (Figure 3D; Supplemental Figure 1F). This same relationship holds true for the inbred lines CML247 and NC358 (Figures 3E and 3F; Supplemental Figures 1G and 1H).

In contrast to the severe lines, three additional lines rescued the Lgn-R phenotype when heterozygous at Sol. Average plant heights and leaf widths for the Sol-Nc350/Sol-B heterozygotes were, respectively, in Albany 141.6 ± 14.22 and 6.7 ± 0.7 cm and in Davis 72.0 ± 23.5 and 3.5 ± 1.0 cm. These averages are all greater than their Sol-B/Sol-B counterparts, which measured 116.4 ± 13.4 and 5.3 ± 0.7 cm in Albany and 52.5 ± 14.7 and 1.75 ± 0.5 cm in Davis. A one-way ANOVA followed by a Tukey’s posthoc test (95% CI) found significant effects from genotype (P = 2e-16), location (P = 0.0002), and genotype-by-location interaction (P = 1.6e-11) for the family (Figure 3G; Supplemental Figure 1I). These same measurements for Sol-Tzi8/Sol-B heterozygotes were 158.5 ± 15.4 and 6.5 ± 0.4 cm in Albany and were significantly increased over their Sol-B/Sol-B relatives, which measured 120.4 ± 12.5 and 4.9 ± 0.4 cm, with the same trends in Davis. A one-way ANOVA followed by a Tukey’s posthoc test (95% CI) found significant effects from genotype (P = 1.1e-11), location (P = 6.4e-13), and genotype-by-location interaction (P = 5.0e-11) for the family (Figure 3H; Supplemental Figure 1J). The third rescuing line, M162W, showed a similar pattern (Figure 3I; Supplemental Figure 1K).

Analysis of single-nucleotide polymorphism (SNP) data from Panzea (Zhao et al., 2006) for the genes in the Sol mapping interval revealed coding sequence variation in GRMZM2G075262 that suggests that this gene is responsible for rescuing Lgn-R phenotypic defects. Of the four genes, the GRMZM2G049211 alleles were identical, GRMZM2G072892 had a single synonymous SNP, GRMZM2G119850 had 4 nonsynonymous SNPs and 18 synonymous SNPs, and GRMZM2G075262 had the most differences, with 12 nonsynonymous SNPs, 26 synonymous SNPs, and 8 insertions/deletions (indels). The indels were discovered upon direct sequencing of the coding region of this gene in Mo17, B73, and the inbred lines used in crosses. We then asked whether sequence variation correlated with a line’s ability to rescue Lgn-R. We found no informative correlations based on the SNPs in GRMZM2G119850. By contrast, 6 of the 12 nonsynonymous SNPs in the GRMZM2G075262 alleles correlated with the ability of Sol to rescue Lgn-R and 4 of the 7 indels correlated with rescuing ability. Indeed, lines that could rescue Lgn-R had the Mo17 version of these SNPs and indels, while the lines that could not rescue had the B73 versions (Figure 4A; Supplemental Figure 2A). CML228 and CML247 are exceptions to this pattern: they have Mo17-like sequences but do not rescue Lgn-R when heterozygous with B73 at Sol.

Figure 4.

Sequence Analysis Supports GRMZM2G075262 as Sol.

(A) Partial alignment of SOL sequence in rescuing and nonrescuing maize inbred lines. The noncanonical ZF motif is in a purple box. Amino acid (AA) substitutions that segregate with ability to rescue are in orange boxes. Indels that segregate with ability to rescue are in red boxes. The second ZF domain is labeled on the cartoon, and the site of an EMS-induced point mutation, Sol-G489E, is marked on the cartoon with a blue line.

(B) Normalized fold expression of syntenic genes in the Sol QTL interval according to background. All treatments represent at least three biological replicates.

(C) RT-qPCR results examining Sol expression in mutant and nonmutant siblings in severe (CML228 and Ms71) and rescued (Nc350) backgrounds. Two-tailed t tests found Sol expression to be statistically higher only in the severe mutant backgrounds [CML228, t(2) = 29.9, P = 0.001; Ms71, t(4) = 2.3, P = 0.05].

Error bars in (B) and (C) represent se.

(D) Plant heights and leaf widths at maturity in an EMS-mutagenized Lgn-R population. The plant that was found to contain a point mutation in GRMZM2G075262 is shown in red.

(E) SOL protein tree. SOL and EDR4 are indicated with red stars. The tree was generated with MEGA X using the following species: Zea mays, Brachypodium distachyon, Sorghum bicolor, Oryza sativa japonica, Setaria italica, Arabidopsis thaliana, Brassica rapa, Gossypium raimondii, Solanum lycopersicum, Populus trichocarpa. Note that the bottom four branches represent collapsed nodes.

We investigated the expression of the four genes from the Sol QTL interval via RT-qPCR. This analysis revealed that GRMZM2G075262 transcript levels are significantly higher in Lgn-R B73 shoot apices than in Lgn-R NIL and nonmutant siblings. None of the other syntenic genes in the interval had trends in expression levels that directly correlated with severity of the Lgn-R phenotype (Figure 4B).

To investigate why CML228 and CML247 have GRMZM2G075262 sequences similar to Mo17 but fail to rescue, we performed RT-qPCR in populations segregating for Lgn-R in these families. For comparison, we used Ms71, which has the B73 Sol haplotype and results in a severe Lgn-R phenotype, and Nc350, which has the Mo17 Sol haplotype and rescues Lgn-R (Figure 4A). As hypothesized, Sol is induced in Lgn-R plants in the Ms71 background but not in Lgn-R plants in a Sol-Nc350/Sol-B background (Figure 4C). We found that Sol transcripts are increased in lines that are Lgn-R and CML228/Sol-B. Results with CML247 were inconsistent across replicates. This result shows that Sol-Nc350, like Sol-M, is capable of reducing Sol-B expression in the heterozygous condition but that Sol-CML228 is not, despite the similarity in sequence to Sol-M. Thus, allelic differences in protein sequence and in regulation of gene expression may both contribute to variation at Sol.

A mutagenesis screen supported our hypothesis that GRMZM2G075262 is the gene responsible for the effects of the Sol locus on the Lgn-R phenotype. Lgn-R plants that had been backcrossed three times to Mo17 were mutagenized with EMS and crossed onto B73. The resulting 2000 kernels were planted in the field and scored for Lgn-R phenotypes. Whereas almost all of the 1000 Lgn-R plants showed partial rescue of plant height and leaf width, we identified one plant with a more severe Lgn-R phenotype that, when sequenced at GRMZM2G075262, contained a G489E missense mutation in Sol-M (Figures 4A, blue line, and 4D; Supplemental Figure 2A, blue box). We did not observe any developmental phenotypes of Sol-G489E without Lgn-R in the background.

GRMZM2G075262, henceforth referred to as Sol, encodes a homolog of Arabidopsis EDR4. We used BLASTp to identify similar proteins in representative plant species, including several grasses and eudicots, and constructed a phylogenetic tree to describe the relationship of these proteins. SOL and AtEDR4 are in a monophyletic clade of closely related proteins, confirming that the two are homologs (Figure 4E). AtEDR4 and SOL share 61.36% identity across 42% of the protein, which includes a C-terminal Zn ribbon 12 domain and an N-terminal Zn finger-like domain (Supplemental Figure 2B). The maize genome includes another three EDR4-like genes within the same clade as Sol; however, these have no assigned function and have not been previously investigated. The only EDR4-like protein that has been intensively studied is AtEDR4 (Wu et al., 2015).

We also determined the cellular compartment for SOL-M and SOL-B. Using Nicotiana benthamiana for 35S:Sol-B-YFP and 35S:Sol-M-YFP transient expression, we determined that SOL-B localizes to the cell periphery and nucleus. SOL-M, by contrast, localized only to the cell periphery (Figures 5A to 5F). Although these experiments were performed in N. benthamiana and not maize, they support the idea that SOL-B and SOL-M have functional differences. Our combined data from analysis of Lgn-R in multiple inbreds, gene expression levels in different genetic backgrounds, a revertant screen, and cell localization all strongly support the identity of Sol as GRMZM2G075262, a maize homolog of EDR4.

Figure 5.

Sol Exhibits Expression and Localization Differences That Are Dependent on Background and Treatment.

(A) to (C) SOL-B-YFP localization shown with merged images (A), the YFP filter (B), and the DAPI filter (C).

(D) to (F) SOL-M-YFP localization shown with merged images (D), the YFP filter (E), and the DAPI filter (F).

(G) Sol and Pr4 normalized fold expression as determined by RT-qPCR at four different time points throughout chitin exposure. Two-tailed t tests found Sol expression to be statistically greater in the treatment versus the control at the 10-min [t(5) = 2.59, P = 0.05], 30-min [t(3) = 2.3, P = 0.01], and 60-min [t(5) = 3.4, P = 0.02] time points. Two-tailed t tests found Pr4 induction to be significant at the 30-min [t(3) = 3.2, P = 0.05] and 60-min [t(3) = 3.5, P = 0.04] time points.

(H) Sol and Pr4 induction at 60-min time points in B73 and Lgn-R B73 backgrounds. Two-tailed t tests found significant Sol induction in nonmutant samples treated with flg22 [t(3) = 4.5, P = 0.02], and Lgn-R B73 samples treated with chitin showed significant repression of Sol [t(5) = 2.7, P = 0.04]. Two-tailed t tests also found significant Pr4 induction in nonmutant B73 treated with flg22 [t(4) = 4.3, P = 0.01] and Lgn-R B73 treated with chitin [t(4) = 8.3, P = 0.001] and flg22 [t(3) = 3.4, P = 0.03].

Error bars in (G) and (H) represent se, and data represent a minimum of three biological replicates.

Sol Expression Is Induced by Elicitors of Innate Immunity

AtEDR4 expression shows induction 1 h after treatment with the bacterial elicitors flg22 and HrpZ (Winter et al., 2007). To test if Sol is also transcriptionally induced in response to pathogen elicitors, we measured Sol transcript levels 10, 30, and 60 min and 24 h after treatment with chitin (a fungal elicitor) or water (control). Endochitinase PR4 (PR4), known to be induced in maize by chitin and other elicitors (Zhang et al., 2017), was used to confirm the efficacy of the treatments. By RT-qPCR, we found that Sol was strongly increased in expression within 1 h after treatment with chitin in B73 (wild-type) leaves (Figure 5G). We also treated B73 with flg22 (a bacterial elicitor) and treated Lgn-R B73 tissue with chitin, flg22, or water and collected samples at the 60-min time point. We found that Sol in B73 was also induced by flg22 at the 60-min time point, whereas Sol levels were not statistically induced in Lgn-R after treatment with chitin or flg22 (Figure 5H), presumably because mock-treated Lgn-R leaves already showed significant induction levels compared with the nonmutant B73 controls. In fact, we observed a significant reduction in Sol expression at the 60-min time point in Lgn-R tissue treated with chitin compared with the control. The induction of Sol by elicitors indicates conservation of this particular transcriptional response across species, supporting its identity as a homolog of EDR4.

Lgn-R Putatively Triggers a MAP Kinase Innate Immunity Signaling Cascade

To narrow our search for pathways affected by the Lgn-R mutation, we used a transcriptomic approach to identify differentially expressed genes (DEGs) in Lgn-R versus nonmutant siblings. We found 1568 DEGs (false discovery rate [FDR] < 0.05) in Lgn-R shoot apices compared with nonmutant siblings, of which 1119 were induced and 448 were repressed. A large number of the Lgn-R DEGs are related to disease resistance, including induction of genes that encode receptor-like kinases, WRKY transcription factors, several classical PATHOGENESIS-RELATED PROTEINS, and nucleotide binding Leu-rich repeat Nod-Like Receptors (NLRs; Figures 6A to 6C; Supplemental Data Set 1). This transcriptional signature suggests that an innate immunity signaling cascade is activated in Lgn-R mutants in the absence of pathogen infection. Therefore, we hypothesize that loss of LGN activity causes an “autoimmune syndrome” with developmental consequences.

Figure 6.

Integrated Quantitative RNA-Seq and Phosphoproteomics Reveal a Possible MAPK Signaling Cascade That Induces Immune Responses in Lgn-R.

(A) to (C) RNA-Seq of Lgn-R and nonmutant siblings revealed hundreds of significantly DEGs. Here, we show changes in transcript abundance as fold change in fragments per kilobase of transcript per million mapped reads in Lgn-R compared with the wild type. Among these genes, the most significantly overrepresented categories are genes encoding WRKY transcription factors (A), PATHOGENESIS-RELATED (PR) proteins that are typically induced when plants detect pathogens (B), and receptors, including several families of receptor-like kinases (RLKs) and Nod-like receptors that are involved in sensing and responding to pathogen infection (C).

(D) LGN resides at the plasma membrane and signals through its kinase activity to promote leaf development and block a MAPK immunity cascade. In the presence of the Lgn-R mutation, leaf development is compromised and the MAPK cascade is activated, leading to a pathogen-triggered immunity (PTI) response that can be dampened by cool temperatures or the presence of Sol-M.

To identify possible substrates of Lgn-R, we took a global phosphoproteomic approach. Three biological replicates of shoot apex tissue were collected from Lgn-R B73 and nonmutant siblings. We detected phosphopeptides in proteins encoded by 3000 different maize genes across all samples. Thirty-five proteins were phosphorylated in all wild-type samples but not in any Lgn-R samples (Supplemental Table 2); these could therefore be peptides that are uniquely phosphorylated by LGN or downstream of LGN signaling. There is no clear pattern among these potential substrates, although these proteins include a number of transcription factors, chromatin-remodeling factors, proteins involved in cell wall synthesis, and cell cycle regulators that could impact Lgn-R phenotypes.

EDR4 is proposed to suppress MAP kinase (MAPK)-mediated immune responses by controlling the localization of a MAP kinase kinase kinase (MAP3K), called EDR1 (or AtMAP3Kδ3), which acts as an upstream repressor of AtMPK3/AtMPK6 activity (Wu et al., 2015). Moreover, AtMPK3/AtMPK6-mediated defense signaling is suppressed at cool temperatures (Cheng et al., 2013), similar to the suppression of Lgn-R phenotypes at cool temperatures. These correlations led us to speculate that Sol could function by interacting with an immunity-related MAPK signaling cascade that is activated in the Lgn-R mutant. To investigate this possibility, we searched for signs of a differentially phosphorylated signaling cascade in the Lgn-R phosphoproteome (compared with the wild-type phosphoproteome), focusing on phosphopeptides identified in our data set that have strong similarity to well-characterized phosphorylation sites on proteins in other plant species. With this targeted approach, we identified a putative signaling cascade that could act downstream of LGN-R to promote the expression of pathogen defense-related genes and disrupt development. Although not statistically significant, we did identify putative differentially phosphorylated sites on homologs of BIR3, BAK1, CDG1, BSU1, SHAGGY, MAP3K (YODA-like), MAP2Ks, MAPKs, and WRKYs that are consistent with activation of this innate immunity MAPK signaling cascade in Lgn-R but not in nonmutant siblings (Supplemental Data Set 2). Alignments of the shared phosphosites in BAK1, BSU1, and SHAGGY-Like BIN2 between maize and Arabidopsis are shown in Supplemental Figures 2C to 2E. Another canonical MAPK substrate, MKP1, is also hyperphosphorylated in Lgn-R, further indicating that LGN regulates MAPK signaling (Park et al., 2011). Thus, we suggest that LGN could act as an upstream repressor of a BAK1-SHAGGY-MAPK signaling network that induces expression of pathogen defense genes.

DISCUSSION

Lgn-R is a semidominant maize mutant with striking leaf defects. The phenotype is severe in B73 and nearly wild type in Mo17. Analysis of a NIL that segregates for Mo17 at the Sol locus showed that Sol-M rescues severe aspects of the Lgn-R phenotype. One copy of Sol-M partially restores the plant height, leaf width, ligule development, ear development, and provides heat stress survival of the Lgn-R mutant. We mapped Sol to a region containing four genes. Subsequent SNP and expression analyses, as well as a mutagenesis screen, strongly support the hypothesis that GRMZM2G075262, a maize homolog of EDR4, is responsible for the Sol phenotypes. Integrated transcriptomics and phosphoproteomics suggest that the Lgn-R mutation triggers a pathogen defense transcriptional program in the absence of pathogen infection. Sol-M, but not Sol-B, rescues many of the development defects and may act in a feedback loop to suppress immune system activation.

Lgn-R resembles other mutants with constitutively activated pathogen defense responses in certain regards. Autoimmune mutants are characterized by dwarfism, upregulation of biotic stress genes, and sensitivity to growth conditions and genetic background (van Wersch et al., 2016). Plant growth is compromised, presumably because more energy is allocated to defense than to growth and development (Huot et al., 2014). Although few maize autoimmune mutants have been described (Hu et al., 1996), Lgn-R appears unique in its developmental leaf defects, such as the missing ligule and narrow leaves. Many autoimmune mutants caused by lesions in NLR genes (or related pathways), including bonsai, chs3-1, chs2-1, and rpp4 in Arabidopsis and Rp1-D21 in maize, have severe phenotypes at low temperatures that are rescued at higher temperatures (Hua et al., 2001; Jambunathan et al., 2001; Zhang et al., 2003; Yang and Hua, 2004; Huang et al., 2010; Yang et al., 2010; Negeri et al., 2013). Lgn-R follows an opposite pattern in terms of temperature expressivity because its phenotypes are most severe at high temperature and rescued at lower temperatures. We hypothesize that this difference in temperature sensitivity is due to activation of distinct pathogen defense signaling networks: NLR-mediated defenses (sometimes called effector-triggered immunity) are commonly active at low temperatures and suppressed at high temperatures, whereas MAPK-mediated defenses (sometimes called pattern-triggered immunity) are commonly active at high temperatures and suppressed at low temperatures (Cheng et al., 2013). The extreme temperature sensitivity and phosphoproteome of Lgn-R suggest that it is a mutation in pattern-triggered immunity. The mechanisms by which temperature influences pathogenicity, NLR activity, and MAPK signaling remain unclear (Hua, 2013; Huot et al., 2017); future studies on the precise molecular functions of LGN under different temperatures could help illuminate how plant defenses are impacted by the environment in agricultural crops.

Lgn-R developmental defects are only partially rescued by the NIL; the ligule is still incomplete and leaves are narrow compared with Lgn-R in Mo17. In the greenhouse, increasing the copy number of Sol-M does not statistically change the height or leaf width of Lgn-R heterozygotes but does restore a normal ligule. Sol-M statistically improves growth of Lgn-R homozygotes in a dosage-dependent manner. One copy increases leaf width and plant height and two copies further improves their growth, but they are still clearly mutants. These results demonstrate the importance of at least one functional copy of LGN for complete rescue by Sol-M.

When Lgn-R was crossed into additional inbred lines, a pattern of amino acid substitutions and indels in Sol became apparent between the rescuing and severe inbred lines. There were, however, two exceptions to this pattern, CML247 and CML228. Further investigation showed that, despite their Mo17-like coding regions, at least one, CML228, had significantly increased mRNA levels in the Lgn-R background when heterozygous with Sol-B. Promoter differences could help to explain this regulatory difference as well as the failure of these alleles to rescue Lgn-R.

In further support of the idea that relative expression levels of Sol could lead to functional differences, EDR4 and Sol are both transiently induced 60 min after cells are exposed to receptor-like kinase elicitors such as flg22. This result suggests that Sol may be a pathogen-response gene and that SOL-M may act in a negative feedback loop to limit constitutive activation of defense-related MAPK signaling cascades. EDR4 functions in repressing immune responses by relocating EDR1, which has a repressive effect on MAPK signaling cascades, to the plasma membrane in regions of hyphal penetration. The EDR4 and EDR1 interaction is dependent on the presence of the first 150 amino acids in EDR4 (Wu et al., 2015). Interestingly, the majority of indels and nonsynonymous mutations between SOL-B and SOL-M are within the first 150 amino acids of the protein. This finding implies that amino acid differences could also be at the heart of the functional differences between the two versions. Additionally, EDR4 is localized at the membrane (Wu et al., 2015), as are SOL-M and SOL-B when transiently expressed in N. benthamiana. Nuclear localization of SOL-B and not SOL-M suggests either protein misregulation, subfunctionalization of SOL-B, or simply heterologous misexpression of the protein. Nonetheless, our results suggest a number of avenues for further investigation, all of which are potentially related. For example, the amino acid differences could lead to different binding partners, which in turn lead to alternate localization and expression levels. Experiments to elucidate potential binding partners of SOL-M versus SOL-B will help unveil functional differences between the two proteins.

We propose that LGN is capable of repressing immune responses through upstream phosphorylation of a biotic defense signal transduction cascade (Figure 6D). LGN may also function to promote leaf development through an independent signaling network. The ligule and auricle defects found in Lgn-R are unique among autoimmune mutants and could be caused by a distinct developmental pathway affected in the Lgn-R mutants. Recent studies have linked MAPK signaling cascades to leaf angle phenotypes in rice (Ning et al., 2011). Perhaps LGN affects multiple MAPK pathways, some leading to autoimmune responses and others leading to distinct developmental events. Sol is transcriptionally downstream of the immune signaling cascade controlled by LGN and becomes transcriptionally induced in the mutant background. We speculate that SOL-M has a similar function to AtEDR4 and is capable of suppressing the immune response. SOL-B is incapable of rescuing the Lgn-R phenotype, perhaps because it cannot suppress this immune response. Although it is unclear how temperature influences these pathways, the observed phenotypic patterns that we see are more closely associated with MAPK-based immune responses than other types of immune signal transduction cascades (Cheng et al., 2013).

The identification of the modifier Sol, with its significant genetic and molecular signatures among the maize inbreds, could guide future breeding efforts to balance pathogen defense with overall growth and yield. Further study of the developmental and immunity-related roles of LGN should reveal new players in the realm of plant resource allocation when under pathogen attack and unravel the complex signaling network underlying interactions between development and defense.

METHODS

Genetic Material and Crossing Schemes

Lgn-R heterozygotes in a maize (Zea mays) B73 background were crossed to the IBM lines (Lee et al., 2002) and NAM founder lines (McMullen et al., 2009) and scored in Indiana (Buescher et al., 2014). The tallest Lgn-R plants were backcrossed to B73 three to six times for mapping and evaluation. In earlier generations, the most rescued Lgn-R plant in each population was blindly used for the cross, but later generations used genotyping at Sol with the primers IDP1489 (Supplemental Table 1). Because Lgn-R is very severe in an Ms71/B73 hybrid, Lgn-R in a B73 background was first crossed to Mo17 two times and then crossed to Ms71 a minimum of two times as well.

EMS mutagenesis was performed on pollen from Lgn-R heterozygotes backcrossed to Mo17 three times and used to pollinate B73 ears. Approximately 2000 kernels were planted in the field and Lgn-R plants were observed. Severe Lgn-R plants were crossed to B73, and nonmutants were self-pollinated to create lines that segregated for Sol-M. Individuals that were homozygous for Sol-M were sequenced using GRMZM2G075262 primers in Supplemental Table 1.

Field and Phenotypic Measurements

Three weeks after planting, plants were numbered and tissue collected for DNA. Plants were scored at 5 weeks for a phenotypic assessment. Plant height was taken at maturity. Leaf width was measured at the midpoint of the blade on the leaf above the ear node. This leaf was consistently the longest and widest leaf for the nonmutants and usually leaf 6 counting down from the top of the plant. When Lgn-R plants did not make ears, the leaf measured was the sixth down from the top. In Davis, many of the Lgn-R plants died prior to maturity. For one season, Lgn-R B73 plants were measured for plant height before they died. The average high temperature for the months of June through August in Davis, California, is 33.1°C and 23.2°C in Albany, California (data from http://www.ncdc.noaa.gov). The GPS coordinates for the Albany and Davis, California, fields are, respectively, (37.887, −122.300) and (38.535, −121.771). One-way ANOVAs and Tukey’s posthoc tests were conducted in R. In a number of families, the amount of dead plants in Davis did not allow for adequate statistical analysis via ANOVA. In these instances, z statistics were calculated according to the following formula:

where p1 and p2 are the proportions for each genotype (i.e., dead plants in genotype/total plants in genotype), p′ is the combined proportion (i.e., all dead plants/total plants analyzed), and n1 and n2 are the number of individuals analyzed per genotype.

Growth Chamber and Greenhouse Conditions

Plants were grown in 16 h of full-spectrum light of 500 μmol m⁻² s⁻¹ and 8 h of dark. In an attempt to mimic field condition temperatures, treatments included the intervals of 15 to 32°C and 11 to 21°C, where high and low temperatures for each condition were matched to light and dark, respectively. A further experiment maintained the plants at a constant 17°C or a constant 30°C. Watering was maintained sufficiently to maintain soil moisture and up to 1 cm standing in each tray immediately after watering. Soil humidity probes were used to verify consistent hydration between temperature conditions.

The greenhouse conditions vary depending on weather, but typically light intensity is 215 μmol m⁻² s⁻¹, temperature ranges from 68 to 78°F, humidity from 44 to 76%, and watering through drip emitters occurs at a rate of 850 mL/d.

Fine-Mapping, Markers, and Numbers of Recombinants

To fine-map Sol, Lgn-R NIL plants were backcrossed to B73 and progeny were grown in Berkeley, California. All plants were measured for leaf width and plant height. In 2012, markers phi109275 at 51,895,132 and bnlg1811 at 70,769,955 were used to find 41 recombinants from a total of 627 plants. In 2013 and 2014, all plants were genotyped regardless of ligule phenotype. Nonmutant plants that were recombinant at Sol were crossed with Lgn-R B73 to assess the phenotype in the next generation. Recombinants that were Lgn-R were crossed to B73 to confirm the phenotype in the next generation. Additional markers were used to refine the positions of recombinants to the region between the markers bnlg2238 and IDP1489. Primer sequences are given in Supplemental Table 1.

Inbred Sequence Analysis

The original SNP data set used to analyze the four syntenic genes in the Sol QTL interval was the Maize HapMapV2 data (Hufford et al., 2012) found at Panzea.org (Zhao et al., 2006). Using the Genotype Search Tool, SNPs were identified from all relevant inbred lines in the interval on chromosome 1 spanning the coding region of all interval genes. B73 RefGen_v3 mapping locations were used. SNPs present in intronic sequences were ignored. The coding region of GRMZM2G07562 was also directly sequenced in all studied inbred lines using primers described in Supplemental Table 1.

Phylogenetic Analysis

Sequences were curated via a BLASTp search against the nonredundant protein sequence collection using the maize SOL sequence. All nonredundant matches with an e value ≤ 0.001 and a minimum of 15% coverage were selected, downloaded, and aligned using MUSCLE (Edgar, 2004). Sequences were manually curated to remove redundant sequences submitted with nonredundant names and any obvious protein fragments that were less than 100 amino acids in length. This alignment (Supplemental Data Set 3) was fed into MEGA X (Kumar et al., 2018), which we used to compute a maximum likelihood tree with 75 bootstrapping replicates. A Jones-Taylor-Thornton model and Nearest Neighbor Interchange ML Heuristic were used while assuming uniform evolutionary rates between sites.

A specialized function of the BLASTp suite was used to calculate percentage identity for a SOL and AtEDR4 alignment. Both sequences were entered individually using the two-sequence option in BLASTp. These sequences were then directly aligned and compared using the BLASTp software (Schäffer et al., 2001).

PAMP Treatment and RT-qPCR Analysis

Leaf 4 tissue of 4-week-old plants was collected, and PAMP or mock treatments were applied according to the method previously described (Zhang et al., 2017). The PAMPs used include flg22 (GenScript) and hexa-N-acetylchitohexaose (Cayman Chemical). Leaf tissue was taken from at least three separate plants per genotype per treatment per time point. Total RNA was extracted from these tissues using Trizol Reagent (Invitrogen) according to the manufacturer’s specifications, and cDNA was generated using the Superscript III Kits with OligoDt (Thermo Fisher Scientific) following the manufacturer’s recommendations. Two-step RT-qPCR (98°C for 0.2 s, 55°C for 0.5 s) was conducted on a Bio-Rad CFX96 Real-Time System using primers for the housekeeping gene Gapdh to normalize the expression of our target genes: Sol and Pr4. All primers are listed in Supplemental Table 1. Expression levels of the analyzed genes were calculated according to the equation E = Peff (−ΔCt), where Peff is the primer set efficiency calculated using LinRegPCR (Ramakers et al., 2003) and ΔCt was calculated by subtracting the cycle threshold (Ct) value of the housekeeping gene from the Ct values of the gene analyzed. Fold changes were calculated between the ratios of the expression levels of PAMP-treated and mock samples, and expression levels were calculated for three biological replicates.

RNA-Seq Material

Seeds of Lgn-R/IBM × B73 (BC5) were sown in large pots in the greenhouse, and 120 plants were genotyped for Sol. Plants that were obviously Lgn-R were noted, and others were genotyped for Lgn-R, resulting in 23 +/+ Sol-B/Sol-B, 15 +/+ Sol-M/Sol-B, 24 Lgn-R/+ Sol-B/Sol-B, and 24 Lgn-R/+ Sol-M/Sol-B. Shoot apex tissue was harvested 4 weeks after planting. To be consistent between genotypes, we made two biological replicates with seven samples from each genotype. The ScriptSeq v2 RNA-Seq library preparation kit was used in accordance with the manufacturer’s guidelines for sequencing with Illumina Hiseq 2000 as SR50. We assessed raw read quality and checked for adapter contamination using fastqc (http://www.bioinformatics.babraham.ac.uk/projects/fastqc) and multiqc (Ewels et al., 2016). Filtered reads were aligned to the maize B73 genome AGPv3.30 using HISAT2 (Kim et al., 2015) and assigned to AGPv3.30 gene models using HTSeq-counts (Anders et al., 2015), using Cyverse Atmosphere cyberinfrastructure (Merchant et al., 2016). Counted reads were tested for differential expression with edgeR using a GLM approach to account for batch effects between repeated experiments and an FDR significance threshold of 0.05 (McCarthy et al., 2012) and differentially expressed fold changes used for additional plots. Gene functional categories were assessed by a parametric test of gene enrichment with AgriGOv2 (Tian et al., 2017). For the RNA-seq analysis, we combined the two Lgn-R data sets (Sol-B/Sol-M and Sol-B/Sol-B) and combined the two nonmutant data sets to increase the number of replicates in the analysis.

Phosphoproteome

Each pool had ∼9 g of shoot tissue isolated from 3-week-old plants segregating for Lgn-R in the B73 background. Two grams of frozen tissue was ground in liquid nitrogen by a mortar/pestle for 15 min to fine powder and then transferred to a 50-mL conical tube. Proteins were precipitated and washed by 50 mL of −20°C acetone three times, then by 50 mL of −20°C methanol with 0.2 mM Na₃VO₄ three times. Protein pellets were aliquoted into four 2-mL Eppendorf tubes per sample and dried in a SpeedVac at 4°C.

Ground tissue powders were suspended in extraction buffer [8 M urea, 100 mM Tris, 5 mM Tris(2-carboxyethyl)phosphine, and phosphatase inhibitors, pH 7]. Proteins were precipitated by adding 4 volumes of cold acetone and incubated at 4°C for 2 h. Samples were centrifuged at 4,000g, 4°C, for 5 min. Supernatant was removed and discarded. Proteins were resuspended in urea extraction buffer and precipitated by cold acetone. Protein pellets were washed by cold methanol with 0.2 mM Na₃VO₄ to further remove nonprotein contaminants. Protein pellets were suspended in extraction buffer [8 M urea, 100 mM Tris, 5 mM Tris(2-carboxyethyl)phosphine, and phosphatase inhibitors, pH 7]. Proteins were first digested with Lys-C (Wako Chemicals, 125-05061) at 37°C for 15 min. Protein solution was diluted eight times to 1 M urea with 100 mM Tris and digested with trypsin (Roche, 03708969001) for 4 h. Reduced Cys residues were alkylated by adding 10 mM iodoacetamide and incubating at 37°C in the dark for 30 min.

Phosphopeptide enrichment was performed using CeO₂ affinity capture. One percent colloidal CeO₂ (Sigma-Aldrich, 289744) was added to the acidified peptide solution (3% [w/v] trifluoroacetic acid, CeO₂:peptide [w/w] = 1:10). After brief vortexing, CeO₂ with captured phosphopeptides was spun down at 1,000g for 1 min. Supernatant was removed, and the CeO₂ pellet was washed with 1 mL of 1% trifluoroacetic acid. Phosphopeptides were eluted by adding eluting buffer [200 mM (NH₄)₂HPO₄, 2 M NH₃·H₂O, and 10 mM EDTA, pH 9.5, same volume as the added 1% colloidal CeO₂] and vortexing briefly. CeO₂ was precipitated by adding 10% formic acid with 100 mM citric acid (same volume as the added 1% colloidal CeO₂) to a final pH of 3. Samples were centrifuged at 16,100g for 1 min. Supernatant containing phosphopeptides was removed for mass spectrometry analysis.

An Agilent 1100 HPLC system was used to deliver a flow rate of 600 nL min⁻¹ to a custom three-phase capillary chromatography column through a splitter. Column phases are a 30-cm-long reverse phase (RP1; 5 μm Zorbax SB-C18, Agilent), 5-cm-long strong cation exchange (SCX; 5 μm PolySulfoethyl, PolyLC), and 20-cm-long reverse phase 2 (RP2; 2.5 μm BEH C18, Waters), with the electrospray tip of the fused silica tubing pulled to a sharp tip (i.d. < 1 μm). Peptide mixtures were loaded onto RP1, and the three sections were joined and mounted on a custom electrospray adapter for online nested elutions. Peptides were eluted from RP1 section to SCX section using a 0 to 80% acetonitrile gradient for 60 min, and then were fractionated by the SCX column section using a series of 18-step salt gradients of ammonium acetate over 20 min, followed by high-resolution reverse phase separation on the RP2 section of the column using an acetonitrile gradient of 0% to 80% for 120 min. One two dimensional-nano-liquid chromatography-tandem mass spectrometry (LC-MS/MS) required 46 h.

Spectra were acquired on LTQ Velos linear ion trap tandem mass spectrometers (Thermo Electron) employing automated, data-dependent acquisition. The mass spectrometer was operated in positive ion mode with a source temperature of 325°C and a spray voltage of 3000 V. The maximum ion injection time is 50 ms for MS and 100 ms for MS/MS. As a final fractionation step, gas phase separation in the ion trap was employed to separate the peptides into three mass classes prior to scanning; the full MS scan range was divided into three smaller scan ranges (400–750, 750–1000, and 1000–1800 D) to improve dynamic range. Each MS scan was followed by five MS/MS scans of the most intense ions from the parent MS scan. A dynamic exclusion of 1 min was used to improve the duty cycle.

Raw data were extracted and searched using Spectrum Mill vB.06 (Agilent Technologies). MS/MS spectra with a sequence tag length of 1 or less were considered to be poor spectra and were discarded. The remaining MS/MS spectra were searched against the maize B73 RefGen_v2 5b Filtered Gene Set downloaded from www.maizesequence.org. The enzyme parameter was limited to full tryptic peptides with a maximum miscleavage of 1. All other search parameters were set to Spectrum Mill’s default settings (carbamidomethylation of Cys, ±2.5 D for precursor ions, ±0.7 D for fragment ions, and a minimum matched peak intensity of 50%). Ox-Met, n-term pyro-Gln, and phosphorylation on Ser, Thr, or Tyr were defined as variable modifications for phosphoproteome data. A maximum of two modifications per peptide was used. A 1:1 concatenated forward-reverse database was constructed to calculate the FDR. The tryptic peptides in the reverse database were compared with the forward database and were shuffled if they matched to any tryptic peptides from the forward database. Common contaminants such as trypsin and keratin were included in the protein database. There are 127,108 protein sequences in the protein database. Peptide cutoff scores were dynamically assigned to each data set to maintain the FDR < 0.1% at the peptide level. Phosphorylation sites were localized to a particular amino acid within a peptide using the variable modification localization score in Agilent’s Spectrum Mill software. Proteins that share common peptides were grouped using principles of parsimony to address the protein database redundancy issue. Thus, proteins within the same group shared the same set or subset of unique peptides. Protein abundance and phosphorylation levels were quantified via spectral counting. MS runs (replicates) were normalized so that the total number of spectral counts was equal for each run. Spectral counts from technical replicates, when present, were then averaged to get the spectral counts for each biological replicate at the protein level.

Transient Expression in Nicotiana benthamiana

To create 35S:Sol-B-YFP and 35S:Sol-M-YFP, the full-length Sol coding sequence was amplified from cDNA of B73 and Mo17 shoots with specific primers (Supplemental Table 1), cloned into pENTR D, and then recombined into pEarleyGate 101 (Earley et al., 2006) using the Gateway technology, according to the manufacturer’s instructions (Invitrogen). The construct was transformed into Agrobacterium tumefaciens strain GV3101. Transient transformation of N. benthamiana was performed as described (Bolduc and Hake, 2009). Forty-eight hours after agroinfiltration, leaves were observed by LSM710 confocal microscopy (Zeiss) with 470-nm excitation and 535-nm emission filters for YFP and 360-nm excitation and 488-nm emission filters for 4′,6-diamidino-2-phenylindole (DAPI) signal. This experiment was repeated three times.

Accession Numbers

Sequence data from this work can be found under the following accession numbers: Lgn, GRMZM2G134382; Sol, GRMZM2G075262; AtEDR4, AT5G05190; EcPR4, GRMZM2G129189; AtEDR1, AT1G08720; ZmBAK1-like, GRMZM2G115420; ZmBSU1-like, GRMZM2G028700; ZmBIN2-like, GRMZM2G043350; ZmMKP1-like, GRMZM2G005350; RNA-seq data, Bioproject ID PRJNA54785.

Supplemental Data

• Supplemental Figure 1. Ninety-five percent CIs for all tukey’s posthoc tests, statistical support for figures 1 to 3.

• Supplemental Figure 2. Expanded alignments of maize and arabidopsis proteins, supporting figures 4 and 6.

• Supplemental Table 1. Primers used in this study.

• Supplemental Table 2. Phosphopeptides in B73 but not in Lgn-R.

• Supplemental Data Set 1. DEGs from RNA-Seq.

• Supplemental Data Set 2. Phosphopeptides supporting a mAPK signaling cascade in Lgn-R.

• Supplemental Data Set 3. Text file of the alignment used for the phylogenetic analysis shown in figure 4E.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• EDR1 Gramene: AT1G08720

• EDR1 Araport: AT1G08720

• BAK1 Gramene: AT4G33430

• BAK1 Araport: AT4G33430

• BSU1 Gramene: AT1G03445

• BSU1 Araport: AT1G03445

• CDG1 Gramene: AT3G26940

• CDG1 Araport: AT3G26940

• mkp1 Gramene: AT3G55270

• mkp1 Araport: AT3G55270

• PR4 Gramene: AT3G04720

• PR4 Araport: AT3G04720

• BIN2 Gramene: AT4G18710

• BIN2 Araport: AT4G18710

• RPP4 Gramene: AT4G16860

• RPP4 Araport: AT4G16860