Designing rice panicle architecture via developmental regulatory genes

Abstract

Rice panicle architecture displays remarkable diversity in branch number, branch length, and grain arrangement; however, much remains unknown about how such diversity in patterns is generated. Although several genes related to panicle branch number and panicle length have been identified, how panicle branch number and panicle length are coordinately regulated is unclear. Here, we show that panicle length and panicle branch number are independently regulated by the genes Prl5/OsGA20ox4, Pbl6/APO1, and Gn1a/OsCKX2. We produced near-isogenic lines (NILs) in the Koshihikari genetic background harboring the elite alleles for Prl5, regulating panicle rachis length; Pbl6, regulating primary branch length; and Gn1a, regulating panicle branching in various combinations. A pyramiding line carrying Prl5, Pbl6, and Gn1a showed increased panicle length and branching without any trade-off relationship between branch length or number. We successfully produced various arrangement patterns of grains by changing the combination of alleles at these three loci. Improvement of panicle architecture raised yield without associated negative effects on yield-related traits except for panicle number. Three-dimensional (3D) analyses by X-ray computed tomography (CT) of panicles revealed that differences in panicle architecture affect grain filling. Importantly, we determined that Prl5 improves grain filling without affecting grain number.

Keywords: grain filling; near-isogenic lines; panicle architecture; panicle development; rice.

Fig. 1.

Graphical genotypes of near-isogenic lines. (A) Schematic diagrams of the parental lines Koshihikari (chromosomes shown in gray), ST-1 (chromosomes shown in blue), and their derived near-isogenic lines (NILs). Red circles indicate the genomic coordinates of Gn1a, Prl5, and Pbl6. (B) Gross morphologies of Koshihikari, NIL-Prl5^ST-1 + Pbl6^ST-1, and NIL-Gn1a^ST-1 + Prl5^ST-1 + Pbl6^ST-1. Scale bar, 10 cm. (C) Representative panicles of Koshihikari, NIL-Prl5^ST-1 + Pbl6^ST-1, and NIL-Gn1a^ST-1 + Prl5^ST-1 + Pbl6^ST-1. Scale bar, 5 cm.

Fig. 2.

Effects of QTL pyramiding on each panicle trait. (A) Representative panicle morphology of Koshihikari, NIL-Prl5^ST-1 + Pbl6^ST-1, and NIL-Gn1a^ST-1 + Prl5^ST-1 + Pbl6^ST-1. Scale bar, 5 cm. (B–H) Panicle traits in Koshihikari and derived NILs. (B) Panicle length. (C) Panicle rachis length. (D) Average length of the lower three primary branches. (E) Average length of the upper three primary branches. (F) Number of primary branches. (G) Number of secondary branches. (H) Grain number. Data are shown as means ± SD (n = 5 plants). *, P < 0.05 by Tukey’s significant difference test.

Fig. 3.

Effects of QTL pyramiding on panicle branching patterns. (A) Comparison of the interval between branches on each primary branch. The y-axis shows the relative position of each primary branch; the primary branch at the tip is scored as position 1. The value on the x-axis is the interval between branches on each primary branch. The values were calculated by dividing primary branch length by the number of secondary rachilla. Solid lines show the regression curve. (B–H) Interval between branches on each primary branch for Koshihikari and derived NILs. The black line and dots indicate Koshihikari; the color line and dots indicate the NIL: NIL-Prl5^ST-1 (B), NIL-Pbl6^ST-1 (C), NIL-Prl5^ST-1 + Pbl6^ST-1 (D), NIL-Gn1a^ST-1 (E), NIL-Prl5^ST-1 + Gn1a^ST-1 (F), NIL-Gn1a^ST-1 + Pbl6^ST-1 (G), and NIL-Gn1a^ST-1 + Prl5^ST-1 + Pbl6^ST-1 (H). All data in B–H are from n = 5–9 plants.

Fig. 4.

Effects of QTL pyramiding on yield-related traits. (A–D) Yield-related traits in Koshihikari and derived NILs: (A) culm length, (B) panicle number per plant, (C) main panicle weight, and (D) panicle weight per plant. Data are shown as means ± SD (n = 10 plants in A, B, and D, n = 5 plants in C). *, P < 0.05 by Tukey’s significant difference test.

Fig. 5.

Effects of QTL pyramiding on grain traits. (A) Representative grains of Koshihikari and derived NILs NIL-Prl5^ST-1 + Pbl6^ST-1 and NIL-Gn1a^ST-1 + Prl5^ST-1 + Pbl6^ST-1. Scale bars, 1 cm. (B) Thousand-grain weight (g) in Koshihikari and derived NILs. Data are shown as means ± SD (n = 10 plants). (C) Grain volumes, as analyzed by 3D X-ray CT, in Koshihikari, NIL-Gn1a^ST-1 + Pbl6^ST-1, and NIL-Gn1a^ST-1 + Prl5^ST-1 + Pbl6^ST-1. n = 3 plants. The volumes of all grains for all three individuals per strain are plotted. *, P < 0.05 by Tukey’s significant difference test.

Fig. 6.

Three-dimensional landscape of panicle architecture determined by the alleles harbored at Prl5, Pbl6, and Gn1a.