Fig 1.
Phenotypic plasticity of floral identities of eg1.
(a) Morphology and plasticity analysis of six variable phenotypes of eg1 spikelets. Floral structures and two-factor plots with percentages per panicle in the y axis of all variable phenotypes are shown in the left and middle columns, respectively. Effects of G, E and GxE on each variable phenotype shown in the pie charts (right column) were analyzed by two-way ANOVA. Bar = 2 mm in floral structures. (b) Statistical analysis of the phenotypic plasticity of eg1 and wild-type spikelets in two different planting locations. Beijing, Beijing summer. Lingshui, Lingshui winter. (c) Statistical analysis of the phenotypic plasticity of eg1 and wild-type spikelet in two different planting seasons. LS Feb, Lingshui Feb. LS Apr, Lingshui Apr. Wl, WT-like; eg, extra glume; pl, palea to lemma; sp, smaller pa; le, long empty glumes; rs, reiterated spikelets. Values are means ± SE, number of analyzed panicles ≥ 20 in (b), ≥ 10 in (c). Significant difference was determined by ANOVA, *P < 0.05, **P < 0.01.
Fig 2.
Subspecific variations of eg1 floral plasticity.
(a) Floral plasticity of eg1 alleles in two exchanged backgrounds. eg1-1 (ZF802>ZH11) and eg1-2 (ZH11>ZF802) show eg1-1 and eg1-2 backcrossed into ZH11 or ZF802 backgrounds, respectively. (b) Floral plasticity of eg1-4 in indica Dular background. Statistical analysis of two types of panicles according to rs (Type I and Type II) are shown. (c) Floral plasticity of eg1-5 and -6 in japonica Nipponbare background. Statistical analysis of the two independent lines are shown. LS Feb, Lingshui Feb. LS Apr, Lingshui Apr. Beijing, Beijing summer. Lingshui, Lingshui winter. Variable phenotypes of spikelets are defined as in S1 Table, and percentages of them in a panicle are shown in the y axis. le, lemma; pa, palea; st, stamen; eg, empty glume; if, inflorescence primordia; sp, smaller pa; lel, lemma-like organ; pl, palea-lemma mosaic organ. Bars = 2 mm. Values are means ± SE, number of analyzed panicles ≥ 5, and significant difference was determined by ANOVA, *P < 0.05, **P < 0.01.
Fig 3.
Temperature-dependent floral plasticity of eg1.
(a) Floral plasticity of eg1-1 (ZF802) under two different temperature conditions. (b) Floral plasticity of eg1-2 (ZH11) under two different temperature conditions. Spikelet phenotypes and statistical analysis are shown in the left and right of (a) and (b). Variable phenotypes of spikelets are defined as in S1 Table. le, lemma; pa, palea; eg, empty glume; if, inflorescence primordia; lel, lemma-like organ; pl, palea-lemma mosaic organ; leg, long empty glume in spikelet structures. Bars = 2 mm. Values are means ± SE, number of analyzed panicles >10, and significant difference was determined by ANOVA, *P < 0.05, **P < 0.01.
Fig 4.
EG1 encodes a functional lipase predominately localized in mitochondria.
(a) Co-localization of EG1-GFP or GFP-EG1 fusion protein with mitochondria in rice protoplasts. Mitochondria are marked by dye Mito Tracker Red (MT Red) or MTS-mOrange protein. (b) Localization analysis of EG1-GFP and chloroplasts in rice protoplasts. An EG1-GFP driven by 35S or native promoter is shown in the left and right, respectively. Chloroplasts are detected by its auto-fluorescence. Mitochondria are marked by Mito Tracker Red (MT Red). (c) Localization of mitochondrial (35SPro:COX11-GFP) [60] (up) and cellular (35SPro:GFP) (bottom) controls of rice protoplasts. DIC, pictures photographed by differential interference contrast microscope. Bar = 10 μm. (d) Subcellular fractionation assay. Mit, mitochondria fraction; Chl, chloroplasts fraction; α-FLAG, antibody of FLAG-EG1; α-AOX1/2 and α-COXII, specific antibodies of mitochondrial proteins AOX1/2 and COXII; α-RbcL and α-PsbA, specific antibodies of chloroplast proteins RbcL and PsbA. (e) Lipase activity of EG1 in vitro with P-nPB as a substrate at 30°C. EG1 (Full) and EG1 (Δ45) respectively refer to fusion proteins of full-length or no N-terminal (45 aa) protein of EG1 and SUMO peptide. DGL and pp lipase (porcine pancreatic lipase) were used as positive controls. Values are means ± SE for three independent experiments.
Fig 5.
High temperature-dependent manner of EG1 in floral robustness regulation.
(a) RT-qPCR analysis of EG1 expression induced by high temperatures in two wild-types. Values are means ± SE (n = 3), and significant difference was determined by ANOVA, *P < 0.05, **P < 0.01, and rice α-TUBULIN as the reference. (b) Western blot analysis of FLAG-EG1 protein accumulation under different temperatures and different tissues in the EG1 complementation lines for 24 hr. Cp, Complementation lines; Ct, non-transgenic wild-type control. HC, Heavy chain of IgG; NS, Nonspecific band (as a loading control). (c) Temperature-dependent lipase activity of EG1. EG1 (Full) and EG1 (Δ45) respectively refer to full-length and no N-terminal (45 AA) protein of EG1 fused to SUMO peptide. Values are means ± SE for three independent experiments. (d) Floral phenotypes of eg1 mutants in a condition of 40°C light 12 hr / 30°C dark 12 hr. Spikelets of eg1-1 with pl, eg and rs phenotypes are shown on i, ii and iii, respectively. Spikelets of eg1-2 with eg and pl phenotypes are shown on iv and v, and with multilayer lemma-like glumes (lel) and/or undetermined inflorescences primordia are on vi to x. x is the inside of ix. le, lemma; pl, palea-lemma mosaic organ; eg, empty glume; lel, lemma-like organ; pa, palea; st, stamen; if, inflorescence primordia. Bars = 2 mm.
Fig 6.
Genotype (G), environment (E) and genotype-environment interaction (GxE)-dependent gene expressional variations of eg1.
(a) Scatterplots of comparisons of environmentally responsive genes between eg1 alleles and wild-types. x and y axes are values of log2 [ratios of gene expression in Beijing to that in Lingshui] of two genotypes respectively. Points represent wild-type-specific (red), eg1-specific (blue) and shared (green) genes. Dotted lines indicate y = x lines. (b) Comparisons of genes significantly affected by G, E and GxE in wild-type and eg1. Effects were analyzed by two-way ANOVA. (c) Triangular scatterplots of distributions of genes significantly affected by G, E and GxE in wild-type and eg1. Each dot indicates a gene, and the three vertexes of triangle indicate three factors G, E, GxE respectively. The closer distance between a gene and a vertex means the stronger effect of the factor on the gene. Insets show expressions of some representative genes in ZF802/eg1-1 (blue lines) and ZH11/eg1-2 (red lines) of Beijing (BJ) and Lingshui (LS). Effects were analyzed by two-way ANOVA. (d) Comparisons of thirteen major pathways affected by G, E and GxE in wild-type and eg1. Numbers in boxes indicate -log10 (P-value) of the pathway enrichment tested by Fisher's exact test with Bonferroni correction and blank boxes no statistical significance.
Fig 7.
OsMADS1, OsMADS6 and OsG1 mediate the floral robustness regulation of EG1.
(a) Spikelets of wild-type rice. i-iv show outside (i), inside (ii), SEM (iii) and paraffin transverse section (iv) of wild-type spikelets, respectively. (b) Genetic analysis of eg1-1 and mads1/nsr. Photos show the outside (i), inside (ii-iii), SEM (iv) and paraffin transverse section (v) of eg1-1, mads1/nsr and eg1-1 nsr mutant spikelets, respectively. (c) Genetic analysis of eg1-1 and mads6-5. Photos of mads6-5 and eg1-1 mads6-5 are shown as (b). (d) Genetic analysis of eg1-1 and nsr mads6-5. Photos of nsr mads6-5 and eg1-1 nsr mads6-5 are shown as (b). Arrows on iii of nsr mads6-5 and eg1-1 nsr mads6-5 indicate the inactive growing points. (e) Relative expression levels of OsMADS1, OsMADS6 and OsG1 in wild-type and eg1 inflorescences of different growth stages by RT-qPCR. Results are means ± SE. Rice α-TUBULIN was used as the reference. (f) Genetic analysis of eg1 and g1-ele. Empty glume phenotypes of eg1-2 in Lingshui, g1-ele and eg1-2 g1-ele in Beijing/Lingshui are shown respectively. (g) The phenotypic plasticity of le in eg1-2, g1-ele and eg1-2 g1-ele. Values are means ± SE, number of analyzed panicles ≥ 5. le, lemma; pa, palea; st, stamen; pi, pistil; eg, empty glume; ca, carpel; fm, floral meristem; im, inflorescence meristem; if, inflorescence primordia; lel, lemma-like organ; gl, glume-like organ; lp, lemma-palea mosaic organ; log, lodicule-glume mosaic organ. leg, long empty glume. Bars = 2 mm, 200 μm and 200 μm in spikelet structures, SEMs and paraffin transverse sections respectively, except for 0.5 mm in ii of (d).
Fig 8.
A model for EG1 in regulating floral developmental robustness against temperature fluctuation in rice.
EG1 responses high temperature fluctuation at its transcriptional, posttranscriptional and lipase activity levels, and positively regulates the transcriptions of floral identity genes such as OsMADS1, OsMADS6 and OsG1, which are mediated by a mitochondria-associated lipid metabolism, resulting in floral developmental robustness against temperature fluctuation. Additionally, chloroplasts-localized EG1 has also been shown to regulate the transcription of floral identity gene OsMADS1 through JA signaling pathway [54].