Research Article

Teosinte ligule allele narrows plant architecture and enhances high-density maize yields

See allHide authors and affiliations

Science  16 Aug 2019:
Vol. 365, Issue 6454, pp. 658-664
DOI: 10.1126/science.aax5482

Less space but greater maize yield

To meet increasing demands for food, modern agriculture works with increasingly dense plantings. Tian et al. identified a gene in teosinte, the wild ancestor of maize, and used it to alter maize such that the plant has a narrower architecture that nonetheless allows leaves access to sunlight (see the Perspective by Hake and Richardson). The yield advantage only becomes evident with the high-density plantings characteristic of modern agriculture, perhaps explaining why this gene was not brought into the fold during the previous millennia of maize domestication.

Science, this issue p. 658; see also p. 640

Abstract

Increased planting densities have boosted maize yields. Upright plant architecture facilitates dense planting. Here, we cloned UPA1 (Upright Plant Architecture1) and UPA2, two quantitative trait loci conferring upright plant architecture. UPA2 is controlled by a two-base sequence polymorphism regulating the expression of a B3-domain transcription factor (ZmRAVL1) located 9.5 kilobases downstream. UPA2 exhibits differential binding by DRL1 (DROOPING LEAF1), and DRL1 physically interacts with LG1 (LIGULELESS1) and represses LG1 activation of ZmRAVL1. ZmRAVL1 regulates brd1 (brassinosteroid C-6 oxidase1), which underlies UPA1, altering endogenous brassinosteroid content and leaf angle. The UPA2 allele that reduces leaf angle originated from teosinte, the wild ancestor of maize, and has been lost during maize domestication. Introgressing the wild UPA2 allele into modern hybrids and editing ZmRAVL1 enhance high-density maize yields.

Feeding the ever-increasing world population requires increased crop yield from limited arable lands (1). One solution to this challenge is to grow more plants per unit area to achieve higher productivity. However, dense planting imposes competition for water, nutrients, and light. Breeders of the cereal crop maize (Zea mays ssp. mays) have addressed this challenge by adapting plant architecture to dense planting. Plant architecture with more upright leaves (i.e., smaller leaf angle) decreases mutual shading and sustains light capture for photosynthesis despite increased plant density, thus improving the accumulation of leaf nitrogen for grain filling and increasing grain yield (25). With such adaptations, the planting density of maize has increased from 30,000 plants per hectare in the 1930s to >80,000 plants per hectare in the 2010s (6, 7).

The ligular region, between the maize blade and sheath, consists of the ligule and auricle (8) and establishes leaf angle, which determines the plant’s overall architecture. Mutant studies in maize have identified genes essential for development of the ligular region (913). lg1 and lg2 mutants exhibit erect leaf architecture because of defects in ligule and auricle development (10, 13). Such erect leaf architecture facilitates dense planting (3, 4). However, leaves in lg mutant plants are too erect to be useful in commercial hybrids. We searched for natural alleles that optimize plant architecture and leaf angle for dense planting.

Upright Plant Architecture2 regulates leaf angle through ZmRAVL1

We mapped 12 quantitative trait loci (QTLs) for leaf angle in a population of 866 maize–teosinte BC2S3 recombinant inbred lines derived from a cross between the maize inbred line W22 and the teosinte accession CIMMYT 8759 (Z. mays ssp. parviglumis, hereafter referred to as 8759) (Fig. 1A and table S1). The QTL with the largest effect, UPA2 (Upright Plant Architecture2), located on chromosome 2, was selected for positional cloning. The teosinte allele exhibited reduced leaf angle relative to the maize allele (fig. S1A). To verify the allelic effect, we developed a pair of near-isogenic lines (NILs) for UPA2 (UPA2-NILW22 and UPA2-NIL8759) (Fig. 1B) from a recombinant inbred line that was heterozygous in the target region (fig. S1B). The two NILs differed in leaf angle in upper, middle, and lower leaves (Fig. 1C), indicating a canopy-wide effect of UPA2 in altering plant architecture. We performed scanning electron microscopy and histological analyses for UPA2-NIL8759 and UPA2-NILW22, which exhibited similar size of auricle cells (fig. S2) but differed in auricle expansion. Compared with UPA2-NILW22, UPA2-NIL8759 with upright leaf angle had a narrower ligular band, resulting in reduced auricle size at maturity (Fig. 1, D and E, and fig. S3). Sclerenchymal layers in the ligular region provide mechanical strength for the blade. UPA2-NIL8759 contained more layers of sclerenchyma cells on the adaxial side than did UPA2-NILW22 (Fig. 1, F and G).

Fig. 1 Positional cloning of UPA2.

(A) QTL mapping for middle leaf angle in the maize–teosinte BC2S3 population. LOD, logarithm of odds. UPA2 and UPA1 are the largest- and second-largest leaf angle QTL, respectively. The dashed gray line at LOD of 5 indicates the threshold of claiming significant QTLs. (B) Gross morphologies of UPA2-NILW22 and UPA2-NIL8759. The white arrows indicate the lower, middle, and upper leaves in which leaf angle was scored. Scale bars, 20 cm. (C) Comparison of leaf angle in lower, middle, and upper leaves between UPA2-NILW22 and UPA2-NIL8759. (D) Scanning electron microscopy analysis of the ligular region of UPA2-NILW22 and UPA2-NIL8759. The ligular band and the mature auricle region are indicated by white dashed lines. Scale bars, 3 mm (top) and 500 μm (bottom). (E) Comparison of the width of the ligular band and auricle margin between UPA2-NILW22 and UPA2-NIL8759. (F) Cross-sections of the mature ligular region of UPA2-NILW22 and UPA2-NIL8759. Top shows the abaxial side and bottom shows the adaxial side. Scale bars, 100 μm. (G) Comparison of number of the abaxial sclerenchyma cell layers (left) and the adaxial sclerenchyma cell layers (right) between UPA2-NILW22 and UPA2-NIL8759. (H) Location of UPA2 on maize chromosome 2. CEN, centromere. (I) Fine mapping of UPA2 using an NIL population (n = 3180). The number of recombinants between adjacent markers is indicated below the linkage map. (J) Progeny testing of recombinants delimited UPA2 to a 240-bp noncoding region (red lines). The graphical genotypes of the five critical recombinants are shown on the left. White, black, and gray segments indicate regions homozygous for W22, regions homozygous for 8759, and heterozygous regions, respectively. The bar graphs on the right compare middle leaf angle between homozygous recombinants and homozygous nonrecombinants within each recombinant-derived F3 family. Black and white bars represent homozygous progenies that inherited the 8759 and W22 chromosome from the parental recombinant, respectively. The 240-bp region of UPA2 is located 9540 bp upstream of the start codon (ATG) of GRMZM2G102059 (ZmRAVL1). Pink and gray regions indicate the exon and untranslated regions (UTR), respectively. Values are means ± SD. **P < 0.01 (Student’s t test). N.S., not significant.

To identify the genetic factor controlling UPA2, we generated a NIL population (n = 3180) and performed fine mapping following previously described methods (14, 15). UPA2 was narrowed down to a 240–base pair (bp) noncoding region according to the maize reference sequence (Fig. 1, H to J). GRMZM2G102059, encoding a B3 domain–containing protein homologous to RAVL1 in rice (16) (fig. S4), is located 9540 bp downstream of the 240-bp region of UPA2 (Fig. 1, H to J). We thus named GRMZM2G102059ZmRAVL1.” The 240-bp region of UPA2 may function as a distant cis element to regulate ZmRAVL1 expression. Consistent with this hypothesis, UPA2-NIL8759 with small leaf angle showed lower ZmRAVL1 expression than UPA2-NILW22 with large leaf angle in developing leaves (fig. S5A). ZmRAVL1 is localized in the nucleus (fig. S5B). To validate the function of ZmRAVL1, we down-regulated ZmRAVL1 expression by RNA interference (RNAi) and knocked out ZmRAVL1 using CRISPR-Cas9 (17). In the T1 family of ZmRAVL1-KO#1 and ZmRAVL1-KO#2 carrying homozygous-null mutations (fig. S6A), Cas9-free plants were identified and propagated for phenotypic analysis and field trials. We found that both ZmRAVL1 RNAi and knockout lines exhibited smaller leaf angle in lower, middle, and upper leaves compared with wild-type plants (Fig. 2, A and B, and fig. S6, B to D) because of decreased auricle size and increased adaxial sclerenchyma cells in the ligular region (figs. S7 and S8). By contrast, overexpressing ZmRAVL1 led to larger leaf angle in lower, middle, and upper leaves compared with wild type (Fig. 2C and fig. S9). These results indicate that ZmRAVL1 functions as a positive regulator of maize leaf angle.

Fig. 2 ZmRAVL1 regulates maize leaf angle.

(A) ZmRAVL1-knockout lines (ZmRAVL1-KO#1 and ZmRAVL1-KO#2) exhibited reduced leaf angle in lower, middle, and upper leaves. (B) ZmRAVL1 RNAi lines (ZmRAVL1-RNAi#1 and ZmRAVL1-RNAi#2) showed reduced leaf angle in lower, middle, and upper leaves. (C) ZmRAVL1 overexpression lines (ZmRAVL1-OE#1 and ZmRAVL1-OE#2) exhibited increased leaf angle in lower, middle, and upper leaves. WT, wild-type. Values are means ± SD. **P < 0.01 (Student’s t test).

To identify the causative sequence variant in the 240-bp region of UPA2, we sequenced the 240-bp region in W22 and 8759 and identified four sequence differences, including one single-nucleotide polymorphism (SNP) and three one- or two-base sequence polymorphisms, designated S1 to S4, respectively (fig. S10). We investigated whether these four sequence variants were associated with alteration of conserved regulatory sequences that may cause differential regulation by an upstream regulator. Among the four sequence variants, only S2 (TG/–) flanks a putative C2C2-binding motif (AGTGTG) (Fig. 3A and fig. S10). Maize YABBY genes drl1 (drooping leaf1) and drl2 contain a C2C2 zinc-finger domain at the N terminus (12). Their null mutants exhibited increased leaf angle and displayed a dropping-leaf phenotype (12). We tested whether DRL proteins could bind to the sequence surrounding S2 at UPA2. Because of the high sequence similarity between the DRL1 and DRL2 proteins (12), we selected DRL1 for further analysis. Electrophoretic mobility shift assay (EMSA) detected band shifts, and the probe from 8759 containing the TG nucleotides at S2 exhibited a stronger binding affinity for DRL1 compared with the probe from W22 lacking TG nucleotides (Fig. 3B and fig. S11). We also performed chromatin immunoprecipitation–quantitative polymerase chain reaction (ChiP-qPCR) using flag antibody against flag-tagged DRL1 protein. Enrichment was detected in the fragment containing the putative C2C2-binding motif around S2 (Fig. 3C). These results indicated that the S2 variant in UPA2 was associated with differential binding by DRL1.

Fig. 3 DRL1 binds UPA2 and LG1 binds ZmRAVL1 promoter.

(A) Relative locations of UPA2 and ZmRAVL1. S2 (TG/–) flanks a putative C2C2-binding motif (red box). (B) EMSA shows that S2 was associated with differential binding by DRL1. The biotin-labeled probes are indicated in (A). (C) ChIP-qPCR indicates that DRL1 could bind the sequence surrounding S2 in vivo. The fragments used in ChIP-qPCR are indicated in (A). The F3 fragment contains the putative C2C2-binding motif surrounding S2, whereas the F1 and F2 fragments contain no putative C2C2-binding motif. (D) EMSA indicates that LG1 could bind the GTAC sites in the ZmRAVL1 promoter. Biotin-labeled probes and mutant probes are indicated in (A). (E) ChIP-qPCR identified significant enrichment in the fragment containing the GTAC sites in the ZmRAVL1 promoter. The fragments used in ChIP-qPCR are indicated in (A). The F4 fragment contains the putative SBP-binding motifs, whereas the F5 and F6 fragments contain no GTAC sequences. Fold enrichments in (C) and (E) were calculated relative to input. Values are means ± SD (n = 3 repeats) in (C) and (E). **P < 0.01 (Student’s t test). N.S., not significant.

lg1 and lg2 genes establish the blade–sheath boundary, with lg2 functioning before lg1 (10, 13). To determine the regulatory relationship of ZmRAVL1 with lg1 and lg2, we examined their expression in the developing ligular region in their null mutants (fig. S12). In lg1 and lg2 mutants, ZmRAVL1 expression was repressed (fig. S12, A and B), whereas in the ZmRAVL1-knockout line, the expression of lg1 and lg2 exhibited no alteration (fig. S12C). These results, together with the finding that ZmRAVL1-knockout lines exhibited normal ligular development and only affected auricle expansion, suggested that ZmRAVL1 functions downstream of lg1 and lg2. We next tested whether LG1 could regulate ZmRAVL1 expression. Previous studies have shown that SBP-domain proteins recognize and bind GTAC motifs (18). Motif analysis identified two putative GTAC sites in the ZmRAVL1 promoter (Fig. 3A and fig. S13A). Yeast one-hybrid assay (Y1H) (fig. S13B) and EMSA (Fig. 3D and fig. S13C) showed that LG1 could bind the GTAC sites in the ZmRAVL1 promoter in vitro. Further, ChIP-qPCR identified enrichment in the fragment spanning the GTAC sites in the ZmRAVL1 promoter (Fig. 3E). Taken together, these results indicated that LG1 regulates ZmRAVL1 expression.

Our results showed that LG1 could bind ZmRAVL1 promoter, whereas DRL1 could bind the sequence surrounding S2 at UPA2, a distant cis-regulatory variant located 9.5 kb upstream of ZmRAVL1. To further determine how this circuit was regulated, we tested for an interaction between the LG1 and DRL1 proteins. We performed the yeast two-hybrid assay (Y2H) (Fig. 4A) and the split firefly luciferase complementation assay in tobacco leaf epidermal cells (Fig. 4B). Both assays showed that LG1 and DRL1 interact in vitro and in vivo.

Fig. 4 DRL1 and LG1 together regulate ZmRAVL1 expression.

(A) Y2H indicates that LG1 and DRL1 interact in vitro. (B) The split firefly luciferase complementation assay shows that LG1 and DRL1 interact in vivo. (C) Schematic diagram shows the constructs used in the transient transcriptional activity assays. (D) DRL1 represses LG1 to activate the expression of ZmRAVL1. Values are means ± SD (n = 4 repeats). Different letters denote significant differences (P < 0.05) from a Duncan’s multiple-range test.

To investigate how DRL1 and LG1 together regulate ZmRAVL1 expression, we performed transient expression assays in maize protoplasts, in which the coding sequence of DRL1 and LG1 driven by the 35S promoter were used as effectors, and the UPA2 fragment (240 bp) from W22 or 8759 fused into the upstream of the luciferase (LUC) gene driven by a 1.7-kb ZmRAVL1 promoter was used as the reporter (designated W22 and 8759 reporter, respectively) (Fig. 4C). When cotransforming with an empty effector construct, the 8759 reporter exhibited lower LUC activity than did the W22 reporter (Fig. 4D). Overexpressing LG1 alone induced the LUC activities of both the 8759 and W22 reporters, whereas overexpressing DRL1 alone repressed the LUC activities of the two reporters (Fig. 4D). Coexpression of LG1 with DRL1 repressed LG1-activated LUC activity (Fig. 4D). In these three cases, the 8759 reporter displayed lower LUC activity than did the W22 reporter (Fig. 4D). Thus, DRL1 physically interacts with LG1 to attenuate its effect on the transcription of ZmRAVL1.

Brassinosteroid C-6 oxidase underlies UPA1

We also performed fine mapping for the second-largest leaf angle QTL, UPA1 on chromosome 1 (Fig. 1A and table S1). Different from UPA2, the teosinte allele at UPA1 exhibited increased leaf angle relative to the maize allele (fig. S14). Phenotypic analysis of the two NILs for UPA1 showed that UPA1 also has a canopy-wide effect in altering leaf angles in different leaves (Fig. 5, A and B). The difference in leaf angle is due to the alteration in auricle size and the number of sclerenchyma cell layers on the adaxial side (fig. S15). Through fine mapping, UPA1 was delimited to a 223-kb physical region containing a single gene (Fig. 5, C to E), GRMZM2G103773, encoding brassinosteroid C-6 oxidase (brd1), which catalyzes the final steps of brassinosteroid synthesis (19). We compared the brd1 coding sequence between W22 and 8759 and identified seven SNPs, including two nonsynonymous SNPs, but not in the conserved region (fig. S16A). We compared brd1 expression in the two NILs for UPA1. UPA1-NIL8759 with large leaf angle exhibited higher brd1 expression than UPA1-NILW22 with small leaf angle in developing leaves (fig. S16B). Indeed, a brd1-null mutant exhibits dwarfism and aberration in the blade–auricle border (19). We overexpressed brd1 and found that the brd1 overexpression lines exhibited normal ligular development but increased leaf angle (fig. S17) because of the enlarged auricle and the decreased number of sclerenchyma cells on the adaxial side (fig. S18). These results indicate that brd1 is involved in the regulation of maize leaf angle.

Fig. 5 Positional cloning of UPA1.

(A) Gross morphologies of UPA1-NILW22 and UPA1-NIL8759. The white arrows indicate the lower, middle, and upper leaves in which leaf angle was scored. Scale bar, 20 cm. (B) Comparison of leaf angle in lower, middle, and upper leaves between UPA1-NILW22 and UPA1-NIL8759. (C) Location of UPA1 on maize chromosome 1. CEN, centromere. (D) Fine mapping of UPA1 using an NIL population (n = 2040). The number of recombinants between adjacent markers is indicated below the linkage map. (E) Progeny testing of recombinants delimited UPA1 to a 223-kb physical region containing only one annotated gene, GRMZM2G103773 (brd1). The graphical genotypes of the five representative recombinants are shown on the left. White, black, and gray segments indicate regions homozygous for W22, regions homozygous for 8759, and heterozygous regions, respectively. The bar graphs on the right compare middle leaf angle between homozygous recombinants and homozygous nonrecombinants within each recombinant-derived F3 family. (F) Schematic diagram of the brd1 promoter. Red dots indicate the putative E-box motif. (G) EMSA shows that ZmRAVL1 binds the putative E-box motif in the brd1 promoter. The biotin-labeled probe (P2) is indicated in (F). Unlabeled probes were used in the competition assay. (H) ChIP–qPCR identified significant enrichment in the fragment containing the E-box motif in the ZmRAVL1 promoter. The fragment (F7) used in ChIP-qPCR is indicated in (F). Fold enrichments were calculated relative to input. (I) Schematic diagram shows the effectors and reporter used in the transient transcriptional activity assays. (J) ZmRAVL1 activates brd1 expression. Values are means ± SD in (B), (H), and (J). **P < 0.01 (Student’s t test).

In ZmRAVL1-knockout plants, brd1 expression was down-regulated (fig. S19A), whereas in brd1-overexpressed lines, ZmRAVL1 exhibited no change (fig. S19B). The two NILs for UPA2 differed in brd1 expression (fig. S19C), whereas ZmRAVL1 showed similar expression in the two NILs for UPA1 (fig. S19D). These results suggested that brd1 functions downstream of ZmRAVL1. We next tested whether ZmRAVL1 could regulate brd1 expression. B3 transcription factors bind E-box motif (CANNTG) (16). We identified five putative E-box elements in the brd1 promoter (Fig. 5F). Y1H, EMSA, and ChIP-qPCR showed that ZmRAVL1 could bind the brd1 promoter in vitro and in vivo (Fig. 5, G and H, and fig. S20). Transient expression assay in maize protoplasts confirmed that ZmRAVL1 activates brd1 transcription (Fig. 5, I and J).

Of the various phytosteroids, only brassinolide and its immediate precursor, castasterone, possess biological activity in planta (20). brd1 encodes a C-6 oxidase involving the conversion of 6-deoxocastasterone to castasterone (20). We measured endogenous brassinosteroid accumulation in the developing ligular region of the brd1 overexpression and ZmRAVL1-knockout lines. Brassinolide was not detected in our samples. Plants with brd1 overexpression exhibited increased castasterone content (fig. S21). ZmRAVL1-knockout plants that showed reduced brd1 expression had less castasterone than did wild-type plants (fig. S21). These results suggested that ZmRAVL1 affects endogenous brassinosteroid content.

UPA2-UPA1–associated regulatory model for leaf angle

We have developed a model for how the leaf angle difference between UPA2-NIL8759 and UPA2-NILW22 is regulated (fig. S22). We suggest that UPA2 is controlled by a 2-bp sequence variant (TG/–) located 9.5 kb upstream of ZmRAVL1. Compared with the cultivated W22 allele, the wild 8759 allele containing the TG nucleotides at UPA2 is associated with a higher binding affinity for DRL1. DRL1 physically interacts with LG1, which attenuates the transcriptional activation function of LG1 on ZmRAVL1 expression. Low ZmRAVL1 expression in UPA2-NIL8759 results in less expression of brd1, consequently decreasing endogenous brassinosteroid in the ligular region, reducing auricle expansion, and leading to smaller leaf angle. In UPA2-NILW22, low binding affinity for DRL1 at UPA2 releases LG1 to activate ZmRAVL1 transcription, which further up-regulates brd1 expression, resulting in increased brassinosteroid and leaf angle.

Teosinte UPA2 and editing ZmRAVL1 enhance maize yields

To examine the distribution of S2 at UPA2 in the maize natural population, we genotyped S2 in 508 maize inbred lines (21) and 50 maize landraces (table S2). None of these maize accessions carried the TG allele at S2 (table S3). We further genotyped S2 in 45 teosinte accessions (table S4). Only two (4.4%) teosinte accessions carried the TG allele at S2 (table S3). These results suggested that S2 is a rare variant in teosinte that was lost during maize domestication.

Although the TG allele at S2 that reduces leaf angle was lost during maize domestication, this allele could be exploited for generating upright leaf architecture for dense planting in modern breeding. To test this possibility, we planted UPA2-NILW22 and UPA2-NIL8759 under five different planting densities in two environments (Tieling and Sanya, both in China) in 2017. In both field trials, UPA2-NIL8759 with upright leaf angle out-yielded UPA2-NILW22 under high planting densities (>67,500 and 105,000 plants per hectare in Sanya and Tieling, respectively) (Fig. 6, A and B, and fig. S23, A to F). Consistent with this result, UPA2-NIL8759 exhibited a lower rate of decrease of grain yield per plant than UPA2-NILW22 as planting density increased (Fig. 6A and fig. S23C). To evaluate the value of UPA2 in hybrids, we introgressed the teosinte UPA2 allele conferring upright leaf angle, through repeated backcrossing and molecular marker assisted selection, into the two parents of Nongda108 (HuangC × Xu178), an elite maize hybrid widely planted in China. We analyzed yield in field trials of the improved Nongda108 and original Nongda108 hybrids under three different planting densities in Sanya, China, in 2018 (Fig. 6, C and D, and fig. S23, G and H). Under planting densities of 45,000 and 75,000 plants per hectare, the original Nongda108 and the improved Nongda108 exhibited similar grain yield (Fig. 6, C and D). However, as planting density increased to 105,000 plants per hectare, the improved Nongda108 showed higher grain yield than the original Nongda108 (Fig. 6, C and D). These results suggested that the teosinte UPA2 allele can both alter the architecture of modern maize plants and improve yield potentials under dense planting.

Fig. 6 Teosinte UPA2 allele and ZmRAVL1 edited allele enhanced maize grain yield under high planting densities.

(A and B) Comparison of grain yield per plant (A) and grain yield per hectare (B) between UPA2-NILW22 and UPA2-NIL8759 under different planting densities in the field trial in Sanya, China, in 2017. (C and D) Comparison of grain yield per plant (C) and grain yield per hectare (D) between the original Nongda108 and the improved Nongda108 carrying the 8759 UPA2 allele under different planting densities in the field trial in Sanya, China, in 2018. (E and F) Comparison of grain yield per plant (E) and grain yield per hectare (F) between ZmRAVL1-KO#1 and wild-type under different planting densities in the field trial in Tieling, China, in 2018. Values are means ± SD (n = 3 replications). Different letters denote significant differences (P < 0.05) from Duncan’s multiple-range tests. ha, hectare.

ZmRAVL1 edited lines also displayed reduced leaf angle. We evaluated the grain yield of ZmRAVL1-KO#1 (Cas9-free) under five different planting densities in two locations (Tieling and Sanya) in 2018. Yield from ZmRAVL1-KO#1 was greater than from wild-type plants under high planting densities in both locations (Fig. 6, E and F, and fig. S24). In the ZmRAVL1-KO#1 T1 family, Cas9-positive plants were selected and crossed with six elite maize inbreds. In the resulting F1 plants, the wild ZmRAVL1 allele introduced by inbreds was edited (fig. S25). As a result, the F1 hybrids showed reduced leaf angle compared with the control hybrids (fig. S27). We also crossed ZmRAVL1-RNAi#1 with the six maize inbreds. The resulting F1 hybrids exhibited decreased ZmRAVL1 expression compared with the control hybrids (fig. S26), leading to reduced leaf angle (fig. S27). Therefore, manipulating ZmRAVL1 can generate upright leaf architecture for increased planting density.

Supplementary Materials

science.sciencemag.org/content/365/6454/658/suppl/DC1

Materials and Methods

Figs. S1 to S27

Tables S1 to S9

References (2233)

References and Notes

Acknowledgments: We thank J.F. Doebley (University of Wisconsin) for providing teosinte materials and maize landraces and valuable discussion; X. Yang (China Agricultural University) for providing the maize inbred lines; the Maize Genetics COOP Stock Center for providing the maize-teosinte BC2S3 population and mutants; and S. Yang (China Agricultural University) and L. Tan (China Agricultural University) for valuable discussions. Funding: This research was supported by the National Key Research and Development Program of China (2016YFD0100404), the National Natural Science Foundation of China (91535108), the Recruitment Program of Global Experts, and the Fundamental Research Funds for the Central Universities. Author contributions: J.T., C.W., and F.T. designed the research. J.T. and C.W. performed experiments and analyses. J.T., C.W., J.X., L.W., G.X., W.W., D.L., W.Q., X.H., and Q.C. collected phenotypic data. W.J. contributed to transgenic experiments. J.T., C.W., and F.T. wrote and edited the manuscript. All authors read and approved the final manuscript. Competing interests: A patent application related to this work has been submitted by F.T., J.T., and C.W. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article